nips nips2005 nips2005-176 knowledge-graph by maker-knowledge-mining
Source: pdf
Author: Brian Taba, Kwabena Boahen
Abstract: We demonstrate the first fully hardware implementation of retinotopic self-organization, from photon transduction to neural map formation. A silicon retina transduces patterned illumination into correlated spike trains that drive a population of silicon growth cones to automatically wire a topographic mapping by migrating toward sources of a diffusible guidance cue that is released by postsynaptic spikes. We varied the pattern of illumination to steer growth cones projected by different retinal ganglion cell types to self-organize segregated or coordinated retinotopic maps. 1
Reference: text
sentIndex sentText sentNum sentScore
1 edu Abstract We demonstrate the first fully hardware implementation of retinotopic self-organization, from photon transduction to neural map formation. [sent-3, score-0.236]
2 A silicon retina transduces patterned illumination into correlated spike trains that drive a population of silicon growth cones to automatically wire a topographic mapping by migrating toward sources of a diffusible guidance cue that is released by postsynaptic spikes. [sent-4, score-1.966]
3 We varied the pattern of illumination to steer growth cones projected by different retinal ganglion cell types to self-organize segregated or coordinated retinotopic maps. [sent-5, score-1.305]
4 Although these algorithms are readily simulated in software, previous hardware implementations have required high precision components that are expensive in chip area (e. [sent-8, score-0.186]
5 In this paper, we demonstrate topographic refinement of connections between a silicon retina and the first neuromorphic self-organizing map chip, previously reported in [5], which is based on axon migration in the developing brain. [sent-13, score-0.72]
6 Each neurite is tipped by a motile sensory structure called a growth cone that guides the elongating neurite based on local chemical cues. [sent-15, score-0.842]
7 Growth cones move by continually sprouting and retracting finger-like extensions called filopodia whose dynamics can be biased by diffusible ligands in an activitydependent manner [4]. [sent-16, score-0.476]
8 Based on these observations, we designed and fabricated the Neurotrope1 chip to implement a population of silicon growth cones [5]. [sent-17, score-1.25]
9 edu/boahen target x n(x) source c(=) = a b c d Figure 1: Neurotropic axon guidance. [sent-21, score-0.343]
10 Active source cells (grey) relay spikes down their axons to their growth cones, which excite nearby target cells. [sent-23, score-0.792]
11 Neurotropin spreads laterally, establishing a spatial concentration gradient that is sampled by active growth cones. [sent-27, score-0.522]
12 Active growth cones climb the local neurotropin gradient, translating temporal activity coincidence into spatial position coincidence. [sent-29, score-1.107]
13 rotrope1 directly to a spiking silicon retina to illustrate its applicability to larger neuromorphic systems. [sent-31, score-0.405]
14 In Section 2, we present an algorithm for axon migration under the guidance of a diffusible chemical whose release and uptake is gated by activity. [sent-33, score-0.38]
15 In Section 4, we examine the Neurotrope1 system’s performance on a topographic refinement task when driven by spike trains generated by a silicon retina in response to several types of illumination stimuli. [sent-35, score-0.582]
16 2 Neurotropic axon guidance We model the self-organization of connections between two layers of neurons (Fig. [sent-36, score-0.21]
17 Cells in the source layer innervate cells in the target layer with excitatory axons that are tipped by motile growth cones. [sent-38, score-1.217]
18 Growth cones tow their axons within the target layer as directed by a diffusible guidance factor called neurotropin that they bind from the local extracellular environment. [sent-39, score-1.047]
19 Neurotropin is released by postsynaptically active target cell bodies and bound by presynaptically active growth cones, so the retrograde transfer of neurotropin from a target cell to a source cell measures the temporal coincidence of their spike activities. [sent-40, score-1.69]
20 Growth cones move to maximize their neurotropic uptake, a Hebbian-like learning rule that causes cells that fire at the same time to wire to the same place. [sent-41, score-0.565]
21 To prevent the population of growth cones from attempting to trivially maximize their uptake by all exciting the same target cell, we impose a synaptic density constraint that requires a migrating growth cone to displace any other growth cone occupying its path. [sent-42, score-2.611]
22 To state the model more formally, source cell bodies occupy nodes of a regular twodimensional (2D) lattice embedded in the source layer, while growth cones and target cell bodies occupy nodes on separate 2D lattices that are interleaved in the target layer. [sent-43, score-1.889]
23 We index nodes by their positions in their respective layers, using Greek letters for source layer positions (e. [sent-44, score-0.335]
24 , α ∈ Z2 ) and Roman letters for target layer positions (e. [sent-46, score-0.336]
25 Each source cell α fires spikes at a rate aSC (α) and conveys this presynaptic activity down an axon that elaborates an excitatory arbor in the target layer centered on c(α). [sent-49, score-0.834]
26 In principle, every branch of this arbor is tipped by its own motile growth cone, but to facilitate efficient Silicon retina RAM 0 3 Neurotrope1 1 0 2 4 GC 3 1 4 2 1 0 3 1 GC N 4 2 0 3 2 4 N GC AER GC GC N N AER GC USB Computer µC a b c Figure 2: a. [sent-50, score-0.773]
27 Growth cones (GC) occupy nodes of this lattice and extend filopodia to the adjacent nodes. [sent-56, score-0.543]
28 Neurotropin receptors (black) are located at the tip of each filopodium and at the growth cone body. [sent-57, score-0.721]
29 hardware implementation, we abstract the collection of branch growth cones into a single central growth cone that tows the arbor’s trunk around the target layer, dragging the rest of the arbor with it. [sent-61, score-1.843]
30 The arbor overlaps nearby target cells with a branch density A(x − c(α)) that diminishes with distance x − c(α) from the arbor center. [sent-62, score-0.264]
31 The postsynaptic activity aTC(x) of target cell x is proportional to the linear sum of its excitation. [sent-63, score-0.335]
32 aTC(x) = aSC (α)A(x − c(α)) (1) α Postsynaptically active target cell bodies release neurotropin, which spreads laterally until consumed by constitutive decay processes. [sent-64, score-0.394]
33 The neurotropin n(x ) present at target site x is assembled from contributions from all active release sites. [sent-65, score-0.371]
34 The contribution of each target cell x is proportional to its postsynaptic activity and weighted by a spreading kernel N (x − x ) that is a decreasing function of its distance x − x from the measurement site x. [sent-66, score-0.365]
35 n(x ) = aTC (x)N (x − x ) (2) x A presynaptically active growth cone located at c(α) computes the direction of the local neurotropin gradient by identifying the adjacent lattice node c (α) ∈ C(c(α)) with the most neurotropin, where C(c(α)) includes c(α) and its nearest neighbors. [sent-67, score-1.031]
36 c (α) = arg maxx ∈C(c(α)) n(x ) (3) Once the growth cone has identified c (α), it swaps positions with the growth cone already located at c (α), increasing its own neurotropic uptake while preserving a constant synaptic density. [sent-68, score-1.571]
37 Growth cones compute position updates independently, at a rate λ(α) ∝ aSC (α) maxy∈C(c(α)) n(x ). [sent-69, score-0.436]
38 Software simulation of a similar set of equations generates self-organized feature maps when driven by appropriately correlated source cell activity [6]. [sent-71, score-0.287]
39 Here, we illustrate topographic map formation in hardware using correlated spike trains generated by a silicon retina. [sent-72, score-0.515]
40 Cell bodies tag their spikes with their own source layer addresses, which the forward lookup table translates into target layer destinations. [sent-75, score-0.839]
41 Axon updates are computed by growth cones, which decode their own target layer addresses through the reverse lookup table to obtain the source layer addresses of their cell bodies that identify their entries in the forward lookup table. [sent-77, score-1.709]
42 Growth cones move by modifying their entries in the forward and reverse lookup tables to reroute their spikes to updated locations. [sent-79, score-0.659]
43 3 Neurotrope1 system Our hardware implementation splits the model into three stages: the source layer, the target layer, and the intervening axons (Fig. [sent-80, score-0.309]
44 Any population of spiking neurons can act as a source layer; in this paper we employ the silicon retina of [7]. [sent-82, score-0.458]
45 The target layer is implemented by a full custom VLSI chip that interleaves a 48 × 20 array of growth cone circuits with a 24 × 20 array of target cell circuits. [sent-83, score-1.486]
46 1 Axon updates Axon updates are computed by the Neurotrope1 chip using the transistor circuits described in [5]. [sent-94, score-0.258]
47 The Neurotrope1 chip represents neurotropin as charge spreading through a monolithic transistor channel laid out as a honeycomb lattice. [sent-96, score-0.401]
48 Each growth cone occupies one node of this lattice and extends filopodia to the three adjacent nodes, expressing neurotropin receptors at all four locations (Fig. [sent-97, score-0.977]
49 When a growth cone receives a presynaptic spike, its receptor circuits tap charge from all four nodes onto separate capacitors. [sent-99, score-0.843]
50 The first capacitor voltage to integrate to a threshold resets all of the growth cone’s capacitors and transmits a request off-chip to update the growth cone’s position by swapping locations with the growth cone currently occupying the winning node. [sent-100, score-1.726]
51 Each spike is tagged with the address of its originating cell for transmission off-chip. [sent-103, score-0.22]
52 Between chips, spikes are routed through a forward lookup table that translates their original source layer addresses into their destined target layer addresses Retina color map n 0 n 85 n 25 20 15 10 5 n 25 50 75 a b c d Figure 4: Retinotopic self-organization of ON-center RGCs. [sent-104, score-0.943]
53 Silicon retina color map of ON-center RGC body positions. [sent-106, score-0.241]
54 Target layer color map of growth cone positions for sample n = 0, colored by the retinal positions of their cell bodies. [sent-109, score-1.227]
55 Growth cones projected by the representative RGC and its nearest neighbors are outlined in white. [sent-110, score-0.458]
56 Grey lines denote target layer distances used to compute Φ(n) . [sent-111, score-0.309]
57 An axon entry in this forward lookup table is indexed by the source layer address of its cell body and contains the target layer address of its growth cone. [sent-118, score-1.499]
58 Axon updates are computed by growth cone circuits on the Neurotrope1 chip, encoded as address-events, and sent to the ip2022 for processing. [sent-120, score-0.812]
59 These growth cone addresses are translated through a reverse lookup table into the source layer addresses that index the relevant forward lookup table entries (Fig. [sent-122, score-1.38]
60 Modification of the affected entries in each lookup table completes the axon migration (Fig. [sent-124, score-0.343]
61 4 Retinotopic self-organization We programmed the growth cone population to self-organize retinotopic maps by driving them with correlated spike trains generated by the silicon retina. [sent-126, score-1.233]
62 The silicon retina translates patterned illumination in real-time into spike trains that are fed into the Neurotrope1 chip as presynaptic input from different retinal ganglion cell (RGC) types. [sent-127, score-0.909]
63 An ON-center RGC is excited by a spot of light in the center of its receptive field and inhibited by light in the surrounding annulus, while an OFF-center RGC responds analogously to the absence of light. [sent-128, score-0.179]
64 To generate appropriately correlated RGC spike trains, we illuminated the silicon retina with various mixtures of light and dark spot stimuli. [sent-130, score-0.572]
65 Strongly driven RGCs could fire at nearly 1 kHz, which was the highest mean rate at which the silicon retina could still be tuned to roughly balance ON- and OFF-center RGC excitability. [sent-132, score-0.348]
66 We tracked the evolution of the growth cone population by reading out the contents of the lookup table every five minutes, a sampling interval selected to include enough patch stimuli to allow each of the 48 × 20 possible patches to be activated on average at least once per sample. [sent-133, score-0.885]
67 We first induced retinotopic self-organization within a single RGC cell type by illuminating the silicon retina with a sequence of randomly centered spots of light presented against a grey background, selectively activating only ON-center RGCs. [sent-134, score-0.779]
68 Each of the 960 growth RGC stimulus Rate kHz n 0 n 310 n ON center 1 25 20 x 15 OFF center 1 10 5 n 100 x a b c d 200 300 e Figure 5: Segregation by cell type under separate light and dark spot stimulation. [sent-135, score-0.79]
69 Target layer color maps of RGC growth cones at sample n = 0. [sent-143, score-1.158]
70 Black indicates the absence of a growth cone projected by an RGC of this cell type. [sent-144, score-0.896]
71 cones was randomly assigned to a different ON-center RGC, creating a scrambled map from retina to target layer (Fig. [sent-151, score-0.909]
72 The ON-center RGC growth cone population visibly refined the topography of the nonretinotopic initial state (Fig. [sent-153, score-0.791]
73 The definition of retinotopy is that adjacent RGCs innervate adjacent target cells, so we define Φ(n) for a given RGC to be the average target layer distance separating its growth cone from the growth cones projected by the six adjacent RGCs of the same cell type. [sent-156, score-2.383]
74 We next induced growth cones projected by each cell type to self-organize disjoint topographic maps by illuminating the silicon retina with a sequence of randomly centered light or dark spots presented against a grey background (Fig. [sent-159, score-1.742]
75 Half the growth cones were assigned to ON-center RGCs and the other half were assigned to the corresponding OFF-center RGCs. [sent-161, score-0.889]
76 We seeded the system with a random projection that evenly distributed growth cones of both cell types across the entire target layer (Fig. [sent-162, score-1.346]
77 Since only RGCs of the same cell type were coactive, growth cones segregated into ON- and OFF-center clusters on opposite sides of the target layer (Fig. [sent-164, score-1.371]
78 OFF-center RGCs were slightly more excitable on average than ON-center RGCs, so their growth cones refined their topography more quickly (Fig. [sent-166, score-0.961]
79 5e) and clustered in the right half of the target layer, which was also more excitable due to poor power distribution on the Neurotrope1 chip. [sent-167, score-0.168]
80 Finally, we induced growth cones of both cell types to self-organize coordinated retinotopic maps by illuminating the retina with center-surround stimuli that oscillate radially from light to dark or vice versa (Fig. [sent-168, score-1.499]
81 The light-dark oscillation injected enough coactivity between neighboring ON- and OFF-center RGCs to prevent their growth cones from segregating by cell type into disjoint clusters. [sent-170, score-1.075]
82 Instead, both subpopulations developed and maintained coarse retinotopic maps that cover the entire target layer and are oriented in register with one another, properties sufficient to seed more interesting circuits such as oriented receptive fields [10]. [sent-171, score-0.548]
83 Target layer color maps of RGC growth cones for sample n = 0. [sent-180, score-1.158]
84 ior of nominally identical circuits on the Neurotrope1 chip and the silicon retina. [sent-187, score-0.373]
85 In the silicon retina, the wide variance of the RGC output rates [7] limits both the convergence speed and the final topographic level achieved by the spot-driven growth cone population. [sent-188, score-0.988]
86 Variance in the output rates of neighboring RGCs also distorts the shape of the spot stimulus, eroding the fidelity of the correlation-encoded instructions received by the growth cones. [sent-191, score-0.561]
87 Variability in the Neurotrope1 chip further limits topographic convergence. [sent-192, score-0.208]
88 Migrating growth cones are directed by the local neurotropin landscape, which forms an image of recent presynaptic activity correlations as filtered through the postsynaptic activation of the target cell population. [sent-193, score-1.455]
89 This image is distorted by variations between the properties of individual target cell and neurotropin circuits that are introduced during fabrication. [sent-194, score-0.517]
90 5 Conclusions In this paper, we demonstrated a completely neuromorphic implementation of retinotopic self-organization. [sent-196, score-0.195]
91 The only comparable system was described in [11], which processed silicon retina data offline using a software model of neurotrophic guidance running on a workstation. [sent-198, score-0.432]
92 The novel infrastructure developed to implement virtual axon migration allows silicon growth cones to directly interface with an existing family of AER-compliant devices, enabling a host of multimodal neuromorphic self-organizing applications. [sent-200, score-1.354]
93 In particular, the silicon retina’s ability to translate arbitrary visual stimuli into growth cone-compatible spike trains in real-time opens the door to more ambitious experiments such as using natural video correlations to automatically wire more complicated visual feature maps. [sent-201, score-0.856]
94 [2, 3]), which implemented a global winner-take-all function to induce competition, our silicon growth cones compute their own updates using purely local information about the neurotropin gradient, a cellular approach that scales effortlessly to larger populations. [sent-205, score-1.331]
95 Performance might be improved by supplementing our purely morphogenetic model with additional physiologically-inspired mechanisms to prune outliers and consolidate well-placed growth cones into permanent synapses. [sent-206, score-0.889]
96 Gertler (2003), “Cytoskeletal dynamics and transport in growth cone mobility and axon guidance,” Neuron, vol. [sent-242, score-0.872]
97 Boahen (2003), “Topographic map formation by silicon growth cones,” in: Advances in Neural Information Processing Systems 15 (MIT Press, Cambridge, eds. [sent-247, score-0.749]
98 Boahen (2004), “Optic nerve signals in a neuromorphic chip I: Outer and inner retina models,” IEEE Trans. [sent-269, score-0.346]
99 Miller (1994), “A model for the development of simple cell receptive fields and the ordered arrangement of orientation columns through activity-dependent competition between on- and off-center inputs,” J. [sent-284, score-0.174]
100 Kramer (2002), “Coupling an aVLSI neuromorphic vision chip to a neurotrophic model of synaptic plasticity: the development of topography,” Neural Comp. [sent-299, score-0.212]
wordName wordTfidf (topN-words)
[('growth', 0.49), ('cones', 0.399), ('rgc', 0.319), ('cone', 0.231), ('neurotropin', 0.191), ('silicon', 0.189), ('layer', 0.185), ('rgcs', 0.166), ('retina', 0.159), ('axon', 0.151), ('cell', 0.148), ('chip', 0.13), ('target', 0.124), ('lookup', 0.122), ('retinotopic', 0.115), ('topographic', 0.078), ('grey', 0.072), ('spike', 0.072), ('gc', 0.071), ('spot', 0.071), ('source', 0.068), ('bodies', 0.066), ('guidance', 0.059), ('neuromorphic', 0.057), ('hardware', 0.056), ('wire', 0.055), ('circuits', 0.054), ('arbor', 0.053), ('diffusible', 0.051), ('neurotropic', 0.051), ('occupy', 0.051), ('uptake', 0.051), ('trains', 0.05), ('boahen', 0.044), ('excitable', 0.044), ('migration', 0.044), ('addresses', 0.044), ('maps', 0.044), ('map', 0.042), ('population', 0.042), ('light', 0.041), ('presynaptic', 0.04), ('dark', 0.04), ('color', 0.04), ('asc', 0.038), ('atc', 0.038), ('coactivity', 0.038), ('lopodia', 0.038), ('migrating', 0.038), ('retinotopy', 0.038), ('tipped', 0.038), ('spikes', 0.038), ('axons', 0.038), ('retinal', 0.037), ('updates', 0.037), ('postsynaptic', 0.036), ('aer', 0.036), ('adjacent', 0.035), ('cells', 0.034), ('illumination', 0.034), ('motile', 0.033), ('illuminating', 0.033), ('active', 0.032), ('representative', 0.032), ('chips', 0.031), ('usb', 0.03), ('spreading', 0.03), ('coordinated', 0.03), ('lattice', 0.03), ('khz', 0.028), ('topography', 0.028), ('formation', 0.028), ('nodes', 0.028), ('activity', 0.027), ('projected', 0.027), ('positions', 0.027), ('entries', 0.026), ('forward', 0.026), ('receptive', 0.026), ('move', 0.026), ('effortlessly', 0.025), ('honeycomb', 0.025), ('kohonen', 0.025), ('monolithic', 0.025), ('neurite', 0.025), ('neurotrophic', 0.025), ('occupying', 0.025), ('patterned', 0.025), ('postsynaptically', 0.025), ('segregated', 0.025), ('taba', 0.025), ('black', 0.025), ('translates', 0.025), ('release', 0.024), ('virtual', 0.024), ('implementation', 0.023), ('reverse', 0.022), ('spots', 0.022), ('presynaptically', 0.022), ('innervate', 0.022)]
simIndex simValue paperId paperTitle
same-paper 1 0.99999851 176 nips-2005-Silicon growth cones map silicon retina
Author: Brian Taba, Kwabena Boahen
Abstract: We demonstrate the first fully hardware implementation of retinotopic self-organization, from photon transduction to neural map formation. A silicon retina transduces patterned illumination into correlated spike trains that drive a population of silicon growth cones to automatically wire a topographic mapping by migrating toward sources of a diffusible guidance cue that is released by postsynaptic spikes. We varied the pattern of illumination to steer growth cones projected by different retinal ganglion cell types to self-organize segregated or coordinated retinotopic maps. 1
2 0.15412188 1 nips-2005-AER Building Blocks for Multi-Layer Multi-Chip Neuromorphic Vision Systems
Author: R. Serrano-Gotarredona, M. Oster, P. Lichtsteiner, A. Linares-Barranco, R. Paz-Vicente, F. Gomez-Rodriguez, H. Kolle Riis, T. Delbruck, S. C. Liu, S. Zahnd, A. M. Whatley, R. Douglas, P. Hafliger, G. Jimenez-Moreno, A. Civit, T. Serrano-Gotarredona, A. Acosta-Jimenez, B. Linares-Barranco
Abstract: A 5-layer neuromorphic vision processor whose components communicate spike events asychronously using the address-eventrepresentation (AER) is demonstrated. The system includes a retina chip, two convolution chips, a 2D winner-take-all chip, a delay line chip, a learning classifier chip, and a set of PCBs for computer interfacing and address space remappings. The components use a mixture of analog and digital computation and will learn to classify trajectories of a moving object. A complete experimental setup and measurements results are shown.
3 0.10185814 118 nips-2005-Learning in Silicon: Timing is Everything
Author: John V. Arthur, Kwabena Boahen
Abstract: We describe a neuromorphic chip that uses binary synapses with spike timing-dependent plasticity (STDP) to learn stimulated patterns of activity and to compensate for variability in excitability. Specifically, STDP preferentially potentiates (turns on) synapses that project from excitable neurons, which spike early, to lethargic neurons, which spike late. The additional excitatory synaptic current makes lethargic neurons spike earlier, thereby causing neurons that belong to the same pattern to spike in synchrony. Once learned, an entire pattern can be recalled by stimulating a subset. 1 Variability in Neural Systems Evidence suggests precise spike timing is important in neural coding, specifically, in the hippocampus. The hippocampus uses timing in the spike activity of place cells (in addition to rate) to encode location in space [1]. Place cells employ a phase code: the timing at which a neuron spikes relative to the phase of the inhibitory theta rhythm (5-12Hz) conveys information. As an animal approaches a place cell’s preferred location, the place cell not only increases its spike rate, but also spikes at earlier phases in the theta cycle. To implement a phase code, the theta rhythm is thought to prevent spiking until the input synaptic current exceeds the sum of the neuron threshold and the decreasing inhibition on the downward phase of the cycle [2]. However, even with identical inputs and common theta inhibition, neurons do not spike in synchrony. Variability in excitability spreads the activity in phase. Lethargic neurons (such as those with high thresholds) spike late in the theta cycle, since their input exceeds the sum of the neuron threshold and theta inhibition only after the theta inhibition has had time to decrease. Conversely, excitable neurons (such as those with low thresholds) spike early in the theta cycle. Consequently, variability in excitability translates into variability in timing. We hypothesize that the hippocampus achieves its precise spike timing (about 10ms) through plasticity enhanced phase-coding (PEP). The source of hippocampal timing precision in the presence of variability (and noise) remains unexplained. Synaptic plasticity can compensate for variability in excitability if it increases excitatory synaptic input to neurons in inverse proportion to their excitabilities. Recasting this in a phase-coding framework, we desire a learning rule that increases excitatory synaptic input to neurons directly related to their phases. Neurons that lag require additional synaptic input, whereas neurons that lead 120µm 190µm A B Figure 1: STDP Chip. A The chip has a 16-by-16 array of microcircuits; one microcircuit includes four principal neurons, each with 21 STDP circuits. B The STDP Chip is embedded in a circuit board including DACs, a CPLD, a RAM chip, and a USB chip, which communicates with a PC. require none. The spike timing-dependent plasticity (STDP) observed in the hippocampus satisfies this requirement [3]. It requires repeated pre-before-post spike pairings (within a time window) to potentiate and repeated post-before-pre pairings to depress a synapse. Here we validate our hypothesis with a model implemented in silicon, where variability is as ubiquitous as it is in biology [4]. Section 2 presents our silicon system, including the STDP Chip. Section 3 describes and characterizes the STDP circuit. Section 4 demonstrates that PEP compensates for variability and provides evidence that STDP is the compensation mechanism. Section 5 explores a desirable consequence of PEP: unconventional associative pattern recall. Section 6 discusses the implications of the PEP model, including its benefits and applications in the engineering of neuromorphic systems and in the study of neurobiology. 2 Silicon System We have designed, submitted, and tested a silicon implementation of PEP. The STDP Chip was fabricated through MOSIS in a 1P5M 0.25µm CMOS process, with just under 750,000 transistors in just over 10mm2 of area. It has a 32 by 32 array of excitatory principal neurons commingled with a 16 by 16 array of inhibitory interneurons that are not used here (Figure 1A). Each principal neuron has 21 STDP synapses. The address-event representation (AER) [5] is used to transmit spikes off chip and to receive afferent and recurrent spike input. To configure the STDP Chip as a recurrent network, we embedded it in a circuit board (Figure 1B). The board has five primary components: a CPLD (complex programmable logic device), the STDP Chip, a RAM chip, a USB interface chip, and DACs (digital-to-analog converters). The central component in the system is the CPLD. The CPLD handles AER traffic, mediates communication between devices, and implements recurrent connections by accessing a lookup table, stored in the RAM chip. The USB interface chip provides a bidirectional link with a PC. The DACs control the analog biases in the system, including the leak current, which the PC varies in real-time to create the global inhibitory theta rhythm. The principal neuron consists of a refractory period and calcium-dependent potassium circuit (RCK), a synapse circuit, and a soma circuit (Figure 2A). RCK and the synapse are ISOMA Soma Synapse STDP Presyn. Spike PE LPF A Presyn. Spike Raster AH 0 0.1 Spike probability RCK Postsyn. Spike B 0.05 0.1 0.05 0.1 0.08 0.06 0.04 0.02 0 0 Time(s) Figure 2: Principal neuron. A A simplified schematic is shown, including: the synapse, refractory and calcium-dependent potassium channel (RCK), soma, and axon-hillock (AH) circuits, plus their constituent elements, the pulse extender (PE) and the low-pass filter (LPF). B Spikes (dots) from 81 principal neurons are temporally dispersed, when excited by poisson-like inputs (58Hz) and inhibited by the common 8.3Hz theta rhythm (solid line). The histogram includes spikes from five theta cycles. composed of two reusable blocks: the low-pass filter (LPF) and the pulse extender (PE). The soma is a modified version of the LPF, which receives additional input from an axonhillock circuit (AH). RCK is inhibitory to the neuron. It consists of a PE, which models calcium influx during a spike, and a LPF, which models calcium buffering. When AH fires a spike, a packet of charge is dumped onto a capacitor in the PE. The PE’s output activates until the charge decays away, which takes a few milliseconds. Also, while the PE is active, charge accumulates on the LPF’s capacitor, lowering the LPF’s output voltage. Once the PE deactivates, this charge leaks away as well, but this takes tens of milliseconds because the leak is smaller. The PE’s and the LPF’s inhibitory effects on the soma are both described below in terms of the sum (ISHUNT ) of the currents their output voltages produce in pMOS transistors whose sources are at Vdd (see Figure 2A). Note that, in the absence of spikes, these currents decay exponentially, with a time-constant determined by their respective leaks. The synapse circuit is excitatory to the neuron. It is composed of a PE, which represents the neurotransmitter released into the synaptic cleft, and a LPF, which represents the bound neurotransmitter. The synapse circuit is similar to RCK in structure but differs in function: It is activated not by the principal neuron itself but by the STDP circuits (or directly by afferent spikes that bypass these circuits, i.e., fixed synapses). The synapse’s effect on the soma is also described below in terms of the current (ISYN ) its output voltage produces in a pMOS transistor whose source is at Vdd. The soma circuit is a leaky integrator. It receives excitation from the synapse circuit and shunting inhibition from RCK and has a leak current as well. Its temporal behavior is described by: τ dISOMA ISYN I0 + ISOMA = dt ISHUNT where ISOMA is the current the capacitor’s voltage produces in a pMOS transistor whose source is at Vdd (see Figure 2A). ISHUNT is the sum of the leak, refractory, and calciumdependent potassium currents. These currents also determine the time constant: τ = C Ut κISHUNT , where I0 and κ are transistor parameters and Ut is the thermal voltage. STDP circuit ~LTP SRAM Presynaptic spike A ~LTD Inverse number of pairings Integrator Decay Postsynaptic spike Potentiation 0.1 0.05 0 0.05 0.1 Depression -80 -40 0 Presynaptic spike Postsynaptic spike 40 Spike timing: t pre - t post (ms) 80 B Figure 3: STDP circuit design and characterization. A The circuit is composed of three subcircuits: decay, integrator, and SRAM. B The circuit potentiates when the presynaptic spike precedes the postsynaptic spike and depresses when the postsynaptic spike precedes the presynaptic spike. The soma circuit is connected to an AH, the locus of spike generation. The AH consists of model voltage-dependent sodium and potassium channel populations (modified from [6] by Kai Hynna). It initiates the AER signaling process required to send a spike off chip. To characterize principal neuron variability, we excited 81 neurons with poisson-like 58Hz spike trains (Figure 2B). We made these spike trains poisson-like by starting with a regular 200Hz spike train and dropping spikes randomly, with probability of 0.71. Thus spikes were delivered to neurons that won the coin toss in synchrony every 5ms. However, neurons did not lock onto the input synchrony due to filtering by the synaptic time constant (see Figure 2B). They also received a common inhibitory input at the theta frequency (8.3Hz), via their leak current. Each neuron was prevented from firing more than one spike in a theta cycle by its model calcium-dependent potassium channel population. The principal neurons’ spike times were variable. To quantify the spike variability, we used timing precision, which we define as twice the standard deviation of spike times accumulated from five theta cycles. With an input rate of 58Hz the timing precision was 34ms. 3 STDP Circuit The STDP circuit (related to [7]-[8]), for which the STDP Chip is named, is the most abundant, with 21,504 copies on the chip. This circuit is built from three subcircuits: decay, integrator, and SRAM (Figure 3A). The decay and integrator are used to implement potentiation, and depression, in a symmetric fashion. The SRAM holds the current binary state of the synapse, either potentiated or depressed. For potentiation, the decay remembers the last presynaptic spike. Its capacitor is charged when that spike occurs and discharges linearly thereafter. A postsynaptic spike samples the charge remaining on the capacitor, passes it through an exponential function, and dumps the resultant charge into the integrator. This charge decays linearly thereafter. At the time of the postsynaptic spike, the SRAM, a cross-coupled inverter pair, reads the voltage on the integrator’s capacitor. If it exceeds a threshold, the SRAM switches state from depressed to potentiated (∼LTD goes high and ∼LTP goes low). The depression side of the STDP circuit is exactly symmetric, except that it responds to postsynaptic activation followed by presynaptic activation and switches the SRAM’s state from potentiated to depressed (∼LTP goes high and ∼LTD goes low). When the SRAM is in the potentiated state, the presynaptic 50 After STDP 83 92 100 Timing precision(ms) Before STDP 75 B Before STDP After STDP 40 30 20 10 0 50 60 70 80 90 Input rate(Hz) 100 50 58 67 text A 0.2 0.4 Time(s) 0.6 0.2 0.4 Time(s) 0.6 C Figure 4: Plasticity enhanced phase-coding. A Spike rasters of 81 neurons (9 by 9 cluster) display synchrony over a two-fold range of input rates after STDP. B The degree of enhancement is quantified by timing precision. C Each neuron (center box) sends synapses to (dark gray) and receives synapses from (light gray) twenty-one randomly chosen neighbors up to five nodes away (black indicates both connections). spike activates the principal neuron’s synapse; otherwise the spike has no effect. We characterized the STDP circuit by activating a plastic synapse and a fixed synapse– which elicits a spike at different relative times. We repeated this pairing at 16Hz. We counted the number of pairings required to potentiate (or depress) the synapse. Based on this count, we calculated the efficacy of each pairing as the inverse number of pairings required (Figure 3B). For example, if twenty pairings were required to potentiate the synapse, the efficacy of that pre-before-post time-interval was one twentieth. The efficacy of both potentiation and depression are fit by exponentials with time constants of 11.4ms and 94.9ms, respectively. This behavior is similar to that observed in the hippocampus: potentiation has a shorter time constant and higher maximum efficacy than depression [3]. 4 Recurrent Network We carried out an experiment designed to test the STDP circuit’s ability to compensate for variability in spike timing through PEP. Each neuron received recurrent connections from 21 randomly selected neurons within an 11 by 11 neighborhood centered on itself (see Figure 4C). Conversely, it made recurrent connections to randomly chosen neurons within the same neighborhood. These connections were mediated by STDP circuits, initialized to the depressed state. We chose a 9 by 9 cluster of neurons and delivered spikes at a mean rate of 50 to 100Hz to each one (dropping spikes with a probability of 0.75 to 0.5 from a regular 200Hz train) and provided common theta inhibition as before. We compared the variability in spike timing after five seconds of learning with the initial distribution. Phase coding was enhanced after STDP (Figure 4A). Before STDP, spike timing among neurons was highly variable (except for the very highest input rate). After STDP, variability was virtually eliminated (except for the very lowest input rate). Initially, the variability, characterized by timing precision, was inversely related to the input rate, decreasing from 34 to 13ms. After five seconds of STDP, variability decreased and was largely independent of input rate, remaining below 11ms. Potentiated synapses 25 A Synaptic state after STDP 20 15 10 5 0 B 50 100 150 200 Spiking order 250 Figure 5: Compensating for variability. A Some synapses (dots) become potentiated (light) while others remain depressed (dark) after STDP. B The number of potentiated synapses neurons make (pluses) and receive (circles) is negatively (r = -0.71) and positively (r = 0.76) correlated to their rank in the spiking order, respectively. Comparing the number of potentiated synapses each neuron made or received with its excitability confirmed the PEP hypothesis (i.e., leading neurons provide additional synaptic current to lagging neurons via potentiated recurrent synapses). In this experiment, to eliminate variability due to noise (as opposed to excitability), we provided a 17 by 17 cluster of neurons with a regular 200Hz excitatory input. Theta inhibition was present as before and all synapses were initialized to the depressed state. After 10 seconds of STDP, a large fraction of the synapses were potentiated (Figure 5A). When the number of potentiated synapses each neuron made or received was plotted versus its rank in spiking order (Figure 5B), a clear correlation emerged (r = -0.71 or 0.76, respectively). As expected, neurons that spiked early made more and received fewer potentiated synapses. In contrast, neurons that spiked late made fewer and received more potentiated synapses. 5 Pattern Completion After STDP, we found that the network could recall an entire pattern given a subset, thus the same mechanisms that compensated for variability and noise could also compensate for lack of information. We chose a 9 by 9 cluster of neurons as our pattern and delivered a poisson-like spike train with mean rate of 67Hz to each one as in the first experiment. Theta inhibition was present as before and all synapses were initialized to the depressed state. Before STDP, we stimulated a subset of the pattern and only neurons in that subset spiked (Figure 6A). After five seconds of STDP, we stimulated the same subset again. This time they recruited spikes from other neurons in the pattern, completing it (Figure 6B). Upon varying the fraction of the pattern presented, we found that the fraction recalled increased faster than the fraction presented. We selected subsets of the original pattern randomly, varying the fraction of neurons chosen from 0.1 to 1.0 (ten trials for each). We classified neurons as active if they spiked in the two second period over which we recorded. Thus, we characterized PEP’s pattern-recall performance as a function of the probability that the pattern in question’s neurons are activated (Figure 6C). At a fraction of 0.50 presented, nearly all of the neurons in the pattern are consistently activated (0.91±0.06), showing robust pattern completion. We fitted the recall performance with a sigmoid that reached 0.50 recall fraction with an input fraction of 0.30. No spurious neurons were activated during any trials. Rate(Hz) Rate(Hz) 8 7 7 6 6 5 5 0.6 0.4 2 0.2 0 0 3 3 2 1 1 A 0.8 4 4 Network activity before STDP 1 Fraction of pattern actived 8 0 B Network activity after STDP C 0 0.2 0.4 0.6 0.8 Fraction of pattern stimulated 1 Figure 6: Associative recall. A Before STDP, half of the neurons in a pattern are stimulated; only they are activated. B After STDP, half of the neurons in a pattern are stimulated, and all are activated. C The fraction of the pattern activated grows faster than the fraction stimulated. 6 Discussion Our results demonstrate that PEP successfully compensates for graded variations in our silicon recurrent network using binary (on–off) synapses (in contrast with [8], where weights are graded). While our chip results are encouraging, variability was not eliminated in every case. In the case of the lowest input (50Hz), we see virtually no change (Figure 4A). We suspect the timing remains imprecise because, with such low input, neurons do not spike every theta cycle and, consequently, provide fewer opportunities for the STDP synapses to potentiate. This shortfall illustrates the system’s limits; it can only compensate for variability within certain bounds, and only for activity appropriate to the PEP model. As expected, STDP is the mechanism responsible for PEP. STDP potentiated recurrent synapses from leading neurons to lagging neurons, reducing the disparity among the diverse population of neurons. Even though the STDP circuits are themselves variable, with different efficacies and time constants, when using timing the sign of the weight-change is always correct (data not shown). For this reason, we chose STDP over other more physiological implementations of plasticity, such as membrane-voltage-dependent plasticity (MVDP), which has the capability to learn with graded voltage signals [9], such as those found in active dendrites, providing more computational power [10]. Previously, we investigated a MVDP circuit, which modeled a voltage-dependent NMDAreceptor-gated synapse [11]. It potentiated when the calcium current analog exceeded a threshold, which was designed to occur only during a dendritic action potential. This circuit produced behavior similar to STDP, implying it could be used in PEP. However, it was sensitive to variability in the NMDA and potentiation thresholds, causing a fraction of the population to potentiate anytime the synapse received an input and another fraction to never potentiate, rendering both subpopulations useless. Therefore, the simpler, less biophysical STDP circuit won out over the MVDP circuit: In our system timing is everything. Associative storage and recall naturally emerge in the PEP network when synapses between neurons coactivated by a pattern are potentiated. These synapses allow neurons to recruit their peers when a subset of the pattern is presented, thereby completing the pattern. However, this form of pattern storage and completion differs from Hopfield’s attractor model [12] . Rather than forming symmetric, recurrent neuronal circuits, our recurrent network forms asymmetric circuits in which neurons make connections exclusively to less excitable neurons in the pattern. In both the poisson-like and regular cases (Figures 4 & 5), only about six percent of potentiated connections were reciprocated, as expected by chance. We plan to investigate the storage capacity of this asymmetric form of associative memory. Our system lends itself to modeling brain regions that use precise spike timing, such as the hippocampus. We plan to extend the work presented to store and recall sequences of patterns, as the hippocampus is hypothesized to do. Place cells that represent different locations spike at different phases of the theta cycle, in relation to the distance to their preferred locations. This sequential spiking will allow us to link patterns representing different locations in the order those locations are visited, thereby realizing episodic memory. We propose PEP as a candidate neural mechanism for information coding and storage in the hippocampal system. Observations from the CA1 region of the hippocampus suggest that basal dendrites (which primarily receive excitation from recurrent connections) support submillisecond timing precision, consistent with PEP [13]. We have shown, in a silicon model, PEP’s ability to exploit such fast recurrent connections to sharpen timing precision as well as to associatively store and recall patterns. Acknowledgments We thank Joe Lin for assistance with chip generation. The Office of Naval Research funded this work (Award No. N000140210468). References [1] O’Keefe J. & Recce M.L. (1993). Phase relationship between hippocampal place units and the EEG theta rhythm. Hippocampus 3(3):317-330. [2] Mehta M.R., Lee A.K. & Wilson M.A. (2002) Role of experience and oscillations in transforming a rate code into a temporal code. Nature 417(6890):741-746. [3] Bi G.Q. & Wang H.X. (2002) Temporal asymmetry in spike timing-dependent synaptic plasticity. Physiology & Behavior 77:551-555. [4] Rodriguez-Vazquez, A., Linan, G., Espejo S. & Dominguez-Castro R. (2003) Mismatch-induced trade-offs and scalability of analog preprocessing visual microprocessor chips. Analog Integrated Circuits and Signal Processing 37:73-83. [5] Boahen K.A. (2000) Point-to-point connectivity between neuromorphic chips using address events. IEEE Transactions on Circuits and Systems II 47:416-434. [6] Culurciello E.R., Etienne-Cummings R. & Boahen K.A. (2003) A biomorphic digital image sensor. IEEE Journal of Solid State Circuits 38:281-294. [7] Bofill A., Murray A.F & Thompson D.P. (2005) Citcuits for VLSI Implementation of Temporally Asymmetric Hebbian Learning. In: Advances in Neural Information Processing Systems 14, MIT Press, 2002. [8] Cameron K., Boonsobhak V., Murray A. & Renshaw D. (2005) Spike timing dependent plasticity (STDP) can ameliorate process variations in neuromorphic VLSI. IEEE Transactions on Neural Networks 16(6):1626-1627. [9] Chicca E., Badoni D., Dante V., D’Andreagiovanni M., Salina G., Carota L., Fusi S. & Del Giudice P. (2003) A VLSI recurrent network of integrate-and-fire neurons connected by plastic synapses with long-term memory. IEEE Transaction on Neural Networks 14(5):1297-1307. [10] Poirazi P., & Mel B.W. (2001) Impact of active dendrites and structural plasticity on the memory capacity of neural tissue. Neuron 29(3)779-796. [11] Arthur J.V. & Boahen K. (2004) Recurrently connected silicon neurons with active dendrites for one-shot learning. In: IEEE International Joint Conference on Neural Networks 3, pp.1699-1704. [12] Hopfield J.J. (1984) Neurons with graded response have collective computational properties like those of two-state neurons. Proceedings of the National Academy of Science 81(10):3088-3092. [13] Ariav G., Polsky A. & Schiller J. (2003) Submillisecond precision of the input-output transformation function mediated by fast sodium dendritic spikes in basal dendrites of CA1 pyramidal neurons. Journal of Neuroscience 23(21):7750-7758.
4 0.095831186 188 nips-2005-Temporally changing synaptic plasticity
Author: Minija Tamosiunaite, Bernd Porr, Florentin Wörgötter
Abstract: Recent experimental results suggest that dendritic and back-propagating spikes can influence synaptic plasticity in different ways [1]. In this study we investigate how these signals could temporally interact at dendrites leading to changing plasticity properties at local synapse clusters. Similar to a previous study [2], we employ a differential Hebbian plasticity rule to emulate spike-timing dependent plasticity. We use dendritic (D-) and back-propagating (BP-) spikes as post-synaptic signals in the learning rule and investigate how their interaction will influence plasticity. We will analyze a situation where synapse plasticity characteristics change in the course of time, depending on the type of post-synaptic activity momentarily elicited. Starting with weak synapses, which only elicit local D-spikes, a slow, unspecific growth process is induced. As soon as the soma begins to spike this process is replaced by fast synaptic changes as the consequence of the much stronger and sharper BP-spike, which now dominates the plasticity rule. This way a winner-take-all-mechanism emerges in a two-stage process, enhancing the best-correlated inputs. These results suggest that synaptic plasticity is a temporal changing process by which the computational properties of dendrites or complete neurons can be substantially augmented. 1
5 0.089971915 181 nips-2005-Spiking Inputs to a Winner-take-all Network
Author: Matthias Oster, Shih-Chii Liu
Abstract: Recurrent networks that perform a winner-take-all computation have been studied extensively. Although some of these studies include spiking networks, they consider only analog input rates. We present results of this winner-take-all computation on a network of integrate-and-fire neurons which receives spike trains as inputs. We show how we can configure the connectivity in the network so that the winner is selected after a pre-determined number of input spikes. We discuss spiking inputs with both regular frequencies and Poisson-distributed rates. The robustness of the computation was tested by implementing the winner-take-all network on an analog VLSI array of 64 integrate-and-fire neurons which have an innate variance in their operating parameters. 1
6 0.08712475 8 nips-2005-A Criterion for the Convergence of Learning with Spike Timing Dependent Plasticity
7 0.077331461 164 nips-2005-Representing Part-Whole Relationships in Recurrent Neural Networks
8 0.073478118 28 nips-2005-Analyzing Auditory Neurons by Learning Distance Functions
9 0.065385498 25 nips-2005-An aVLSI Cricket Ear Model
10 0.064363234 7 nips-2005-A Cortically-Plausible Inverse Problem Solving Method Applied to Recognizing Static and Kinematic 3D Objects
11 0.060458016 22 nips-2005-An Analog Visual Pre-Processing Processor Employing Cyclic Line Access in Only-Nearest-Neighbor-Interconnects Architecture
12 0.06010985 134 nips-2005-Neural mechanisms of contrast dependent receptive field size in V1
13 0.058639381 88 nips-2005-Gradient Flow Independent Component Analysis in Micropower VLSI
14 0.058606241 149 nips-2005-Optimal cue selection strategy
15 0.05748491 99 nips-2005-Integrate-and-Fire models with adaptation are good enough
16 0.055861536 129 nips-2005-Modeling Neural Population Spiking Activity with Gibbs Distributions
17 0.052488409 39 nips-2005-Beyond Pair-Based STDP: a Phenomenological Rule for Spike Triplet and Frequency Effects
18 0.051321134 108 nips-2005-Layered Dynamic Textures
19 0.051000234 141 nips-2005-Norepinephrine and Neural Interrupts
20 0.049638089 67 nips-2005-Extracting Dynamical Structure Embedded in Neural Activity
topicId topicWeight
[(0, 0.117), (1, -0.192), (2, -0.049), (3, 0.022), (4, -0.044), (5, 0.083), (6, -0.045), (7, 0.019), (8, 0.015), (9, 0.023), (10, -0.104), (11, 0.018), (12, -0.05), (13, 0.125), (14, -0.115), (15, -0.061), (16, 0.014), (17, -0.055), (18, -0.019), (19, -0.032), (20, 0.06), (21, -0.061), (22, -0.093), (23, 0.051), (24, -0.017), (25, 0.048), (26, 0.012), (27, 0.046), (28, 0.03), (29, -0.013), (30, 0.034), (31, 0.051), (32, 0.012), (33, -0.088), (34, -0.037), (35, -0.043), (36, -0.011), (37, -0.046), (38, 0.117), (39, 0.089), (40, -0.02), (41, 0.039), (42, -0.033), (43, 0.032), (44, -0.01), (45, 0.03), (46, -0.046), (47, 0.0), (48, -0.075), (49, 0.123)]
simIndex simValue paperId paperTitle
same-paper 1 0.97032386 176 nips-2005-Silicon growth cones map silicon retina
Author: Brian Taba, Kwabena Boahen
Abstract: We demonstrate the first fully hardware implementation of retinotopic self-organization, from photon transduction to neural map formation. A silicon retina transduces patterned illumination into correlated spike trains that drive a population of silicon growth cones to automatically wire a topographic mapping by migrating toward sources of a diffusible guidance cue that is released by postsynaptic spikes. We varied the pattern of illumination to steer growth cones projected by different retinal ganglion cell types to self-organize segregated or coordinated retinotopic maps. 1
2 0.71112025 1 nips-2005-AER Building Blocks for Multi-Layer Multi-Chip Neuromorphic Vision Systems
Author: R. Serrano-Gotarredona, M. Oster, P. Lichtsteiner, A. Linares-Barranco, R. Paz-Vicente, F. Gomez-Rodriguez, H. Kolle Riis, T. Delbruck, S. C. Liu, S. Zahnd, A. M. Whatley, R. Douglas, P. Hafliger, G. Jimenez-Moreno, A. Civit, T. Serrano-Gotarredona, A. Acosta-Jimenez, B. Linares-Barranco
Abstract: A 5-layer neuromorphic vision processor whose components communicate spike events asychronously using the address-eventrepresentation (AER) is demonstrated. The system includes a retina chip, two convolution chips, a 2D winner-take-all chip, a delay line chip, a learning classifier chip, and a set of PCBs for computer interfacing and address space remappings. The components use a mixture of analog and digital computation and will learn to classify trajectories of a moving object. A complete experimental setup and measurements results are shown.
Author: Yusuke Nakashita, Yoshio Mita, Tadashi Shibata
Abstract: An analog focal-plane processor having a 128¢128 photodiode array has been developed for directional edge filtering. It can perform 4¢4-pixel kernel convolution for entire pixels only with 256 steps of simple analog processing. Newly developed cyclic line access and row-parallel processing scheme in conjunction with the “only-nearest-neighbor interconnects” architecture has enabled a very simple implementation. A proof-of-concept chip was fabricated in a 0.35- m 2-poly 3-metal CMOS technology and the edge filtering at a rate of 200 frames/sec. has been experimentally demonstrated.
4 0.57807505 25 nips-2005-An aVLSI Cricket Ear Model
Author: Andre V. Schaik, Richard Reeve, Craig Jin, Tara Hamilton
Abstract: Female crickets can locate males by phonotaxis to the mating song they produce. The behaviour and underlying physiology has been studied in some depth showing that the cricket auditory system solves this complex problem in a unique manner. We present an analogue very large scale integrated (aVLSI) circuit model of this process and show that results from testing the circuit agree with simulation and what is known from the behaviour and physiology of the cricket auditory system. The aVLSI circuitry is now being extended to use on a robot along with previously modelled neural circuitry to better understand the complete sensorimotor pathway. 1 In trod u ction Understanding how insects carry out complex sensorimotor tasks can help in the design of simple sensory and robotic systems. Often insect sensors have evolved into intricate filters matched to extract highly specific data from the environment which solves a particular problem directly with little or no need for further processing [1]. Examples include head stabilisation in the fly, which uses vision amongst other senses to estimate self-rotation and thus to stabilise its head in flight, and phonotaxis in the cricket. Because of the narrowness of the cricket body (only a few millimetres), the Interaural Time Difference (ITD) for sounds arriving at the two sides of the head is very small (10–20µs). Even with the tympanal membranes (eardrums) located, as they are, on the forelegs of the cricket, the ITD only reaches about 40µs, which is too low to detect directly from timings of neural spikes. Because the wavelength of the cricket calling song is significantly greater than the width of the cricket body the Interaural Intensity Difference (IID) is also very low. In the absence of ITD or IID information, the cricket uses phase to determine direction. This is possible because the male cricket produces an almost pure tone for its calling song. * School of Electrical and Information Engineering, Institute of Perception, Action and Behaviour. + Figure 1: The cricket auditory system. Four acoustic inputs channel sounds directly or through tracheal tubes onto two tympanal membranes. Sound from contralateral inputs has to pass a (double) central membrane (the medial septum), inducing a phase delay and reduction in gain. The sound transmission from the contralateral tympanum is very weak, making each eardrum effectively a 3 input system. The physics of the cricket auditory system is well understood [2]; the system (see Figure 1) uses a pair of sound receivers with four acoustic inputs, two on the forelegs, which are the external surfaces of the tympana, and two on the body, the prothoracic or acoustic spiracles [3]. The connecting tracheal tubes are such that interference occurs as sounds travel inside the cricket, producing a directional response at the tympana to frequencies near to that of the calling song. The amplitude of vibration of the tympana, and hence the firing rate of the auditory afferent neurons attached to them, vary as a sound source is moved around the cricket and the sounds from the different inputs move in and out of phase. The outputs of the two tympana match when the sound is straight ahead, and the inputs are bilaterally symmetric with respect to the sound source. However, when sound at the calling song frequency is off-centre the phase of signals on the closer side comes better into alignment, and the signal increases on that side, and conversely decreases on the other. It is that crossover of tympanal vibration amplitudes which allows the cricket to track a sound source (see Figure 6 for example). A simplified version of the auditory system using only two acoustic inputs was implemented in hardware [4], and a simple 8-neuron network was all that was required to then direct a robot to carry out phonotaxis towards a species-specific calling song [5]. A simple simulator was also created to model the behaviour of the auditory system of Figure 1 at different frequencies [6]. Data from Michelsen et al. [2] (Figures 5 and 6) were digitised, and used together with average and “typical” values from the paper to choose gains and delays for the simulation. Figure 2 shows the model of the internal auditory system of the cricket from sound arriving at the acoustic inputs through to transmission down auditory receptor fibres. The simulator implements this model up to the summing of the delayed inputs, as well as modelling the external sound transmission. Results from the simulator were used to check the directionality of the system at different frequencies, and to gain a better understanding of its response. It was impractical to check the effect of leg movements or of complex sounds in the simulator due to the necessity of simulating the sound production and transmission. An aVLSI chip was designed to implement the same model, both allowing more complex experiments, such as leg movements to be run, and experiments to be run in the real world. Figure 2: A model of the auditory system of the cricket, used to build the simulator and the aVLSI implementation (shown in boxes). These experiments with the simulator and the circuits are being published in [6] and the reader is referred to those papers for more details. In the present paper we present the details of the circuits used for the aVLSI implementation. 2 Circuits The chip, implementing the aVLSI box in Figure 2, comprises two all-pass delay filters, three gain circuits, a second-order narrow-band band-pass filter, a first-order wide-band band-pass filter, a first-order high-pass filter, as well as supporting circuitry (including reference voltages, currents, etc.). A single aVLSI chip (MOSIS tiny-chip) thus includes half the necessary circuitry to model the complete auditory system of a cricket. The complete model of the auditory system can be obtained by using two appropriately connected chips. Only two all-pass delay filters need to be implemented instead of three as suggested by Figure 2, because it is only the relative delay between the three pathways arriving at the one summing node that counts. The delay circuits were implemented with fully-differential gm-C filters. In order to extend the frequency range of the delay, a first-order all-pass delay circuit was cascaded with a second-order all-pass delay circuit. The resulting addition of the first-order delay and the second-order delay allowed for an approximately flat delay response for a wider bandwidth as the decreased delay around the corner frequency of the first-order filter cancelled with the increased delay of the second-order filter around its resonant frequency. Figure 3 shows the first- and second-order sections of the all-pass delay circuit. Two of these circuits were used and, based on data presented in [2], were designed with delays of 28µs and 62µs, by way of bias current manipulation. The operational transconductance amplifier (OTA) in figure 3 is a standard OTA which includes the common-mode feedback necessary for fully differential designs. The buffers (Figure 3) are simple, cascoded differential pairs. V+ V- II+ V+ V- II+ V+ V- II+ V+ V- II+ V+ V- II+ V+ V- II+ Figure 3: The first-order all-pass delay circuit (left) and the second-order all-pass delay (right). The differential output of the delay circuits is converted into a current which is multiplied by a variable gain implemented as shown in Figure 4. The gain cell includes a differential pair with source degeneration via transistors N4 and N5. The source degeneration improves the linearity of the current. The three gain cells implemented on the aVLSI have default gains of 2, 3 and 0.91 which are set by holding the default input high and appropriately ratioing the bias currents through the value of vbiasp. To correct any on-chip mismatches and/or explore other gain configurations a current splitter cell [7] (p-splitter, figure 4) allows the gain to be programmed by digital means post fabrication. The current splitter takes an input current (Ibias, figure 4) and divides it into branches which recursively halve the current, i.e., the first branch gives ½ Ibias, the second branch ¼ Ibias, the third branch 1/8 Ibias and so on. These currents can be used together with digitally controlled switches as a Digital-to-Analogue converter. By holding default low and setting C5:C0 appropriately, any gain – from 4 to 0.125 – can be set. To save on output pins the program bits (C5:C0) for each of the three gain cells are set via a single 18-bit shift register in bit-serial fashion. Summing the output of the three gain circuits in the current domain simply involves connecting three wires together. Therefore, a natural option for the filters that follow is to use current domain filters. In our case we have chosen to implement log-domain filters using MOS transistors operating in weak inversion. Figure 5 shows the basic building blocks for the filters – the Tau Cell [8] and the multiplier cell – and block diagrams showing how these blocks were connected to create the necessary filtering blocks. The Tau Cell is a log-domain filter which has the firstorder response: I out 1 , = I in sτ + 1 where τ = nC aVT Ia and n = the slope factor, VT = thermal voltage, Ca = capacitance, and Ia = bias current. In figure 5, the input currents to the Tau Cell, Imult and A*Ia, are only used when building a second-order filter. The multiplier cell is simply a translinear loop where: I out1 ∗ I mult = I out 2 ∗ AI a or Imult = AIaIout2/Iout1. The configurations of the Tau Cell to get particular responses are covered in [8] along with the corresponding equations. The high frequency filter of Figure 2 is implemented by the high-pass filter in Figure 5 with a corner frequency of 17kHz. The low frequency filter, however, is divided into two parts since the biological filter’s response (see for example Figure 3A in [9]) separates well into a narrow second-order band-pass filter with a 10kHz resonant frequency and a wide band-pass filter made from a first-order high-pass filter with a 3kHz corner frequency followed by a first-order low-pass filter with a 12kHz corner frequency. These filters are then added together to reproduce the biological filter. The filters’ responses can be adjusted post fabrication via their bias currents. This allows for compensation due to processing and matching errors. Figure 4: The Gain Cell above is used to convert the differential voltage input from the delay cells into a single-ended current output. The gain of each cell is controllable via a programmable current cell (p_splitter). An on-chip bias generator [7] was used to create all the necessary current biases on the chip. All the main blocks (delays, gain cells and filters), however, can have their on-chip bias currents overridden through external pins on the chip. The chip was fabricated using the MOSIS AMI 1.6µm technology and designed using the Cadence Custom IC Design Tools (5.0.33). 3 Methods The chip was tested using sound generated on a computer and played through a soundcard to the chip. Responses from the chip were recorded by an oscilloscope, and uploaded back to the computer on completion. Given that the output from the chip and the gain circuits is a current, an external current-sense circuit built with discrete components was used to enable the output to be probed by the oscilloscope. Figure 5: The circuit diagrams for the log-domain filter building blocks – The Tau Cell and The Multiplier – along with the block diagrams for the three filters used in the aVLSI model. Initial experiments were performed to tune the delays and gains. After that, recordings were taken of the directional frequency responses. Sounds were generated by computer for each chip input to simulate moving the forelegs by delaying the sound by the appropriate amount of time; this was a much simpler solution than using microphones and moving them using motors. 4 Results The aVLSI chip was tested to measure its gains and delays, which were successfully tuned to the appropriate values. The chip was then compared with the simulation to check that it was faithfully modelling the system. A result of this test at 4kHz (approximately the cricket calling-song frequency) is shown in Figure 6. Apart from a drop in amplitude of the signal, the response of the circuit was very similar to that of the simulator. The differences were expected because the aVLSI circuit has to deal with real-world noise, whereas the simulated version has perfect signals. Examples of the gain versus frequency response of the two log-domain band-pass filters are shown in Figure 7. Note that the narrow-band filter peaks at 6kHz, which is significantly above the mating song frequency of the cricket which is around 4.5kHz. This is not a mistake, but is observed in real crickets as well. As stated in the introduction, a range of further testing results with both the circuit and the simulator are being published in [6]. 5 D i s c u s s i on The aVLSI auditory sensor in this research models the hearing of the field cricket Gryllus bimaculatus. It is a more faithful model of the cricket auditory system than was previously built in [4], reproducing all the acoustic inputs, as well as the responses to frequencies of both the co specific calling song and bat echolocation chirps. It also generates outputs corresponding to the two sets of behaviourally relevant auditory receptor fibres. Results showed that it matched the biological data well, though there were some inconsistencies due to an error in the specification that will be addressed in a future iteration of the design. A more complete implementation across all frequencies was impractical because of complexity and size issues as well as serving no clear behavioural purpose. Figure 6: Vibration amplitude of the left (dotted) and right (solid) virtual tympana measured in decibels in response to a 4kHz tone in simulation (left) and on the aVLSI chip (right). The plot shows the amplitude of the tympanal responses as the sound source is rotated around the cricket. Figure 7: Frequency-Gain curves for the narrow-band and wide-band bandpass filters. The long-term aim of this work is to better understand simple sensorimotor control loops in crickets and other insects. The next step is to mount this circuitry on a robot to carry out behavioural experiments, which we will compare with existing and new behavioural data (such as that in [10]). This will allow us to refine our models of the neural circuitry involved. Modelling the sensory afferent neurons in hardware is necessary in order to reduce processor load on our robot, so the next revision will include these either onboard, or on a companion chip as we have done before [11]. We will also move both sides of the auditory system onto a single chip to conserve space on the robot. It is our belief and experience that, as a result of this intelligent pre-processing carried out at the sensor level, the neural circuits necessary to accurately model the behaviour will remain simple. Acknowledgments The authors thank the Institute of Neuromorphic Engineering and the UK Biotechnology and Biological Sciences Research Council for funding the research in this paper. References [1] R. Wehner. Matched filters – neural models of the external world. J Comp Physiol A, 161: 511–531, 1987. [2] A. Michelsen, A. V. Popov, and B. Lewis. Physics of directional hearing in the cricket Gryllus bimaculatus. Journal of Comparative Physiology A, 175:153–164, 1994. [3] A. Michelsen. The tuned cricket. News Physiol. Sci., 13:32–38, 1998. [4] H. H. Lund, B. Webb, and J. Hallam. A robot attracted to the cricket species Gryllus bimaculatus. In P. Husbands and I. Harvey, editors, Proceedings of 4th European Conference on Artificial Life, pages 246–255. MIT Press/Bradford Books, MA., 1997. [5] R Reeve and B. Webb. New neural circuits for robot phonotaxis. Phil. Trans. R. Soc. Lond. A, 361:2245–2266, August 2003. [6] R. Reeve, A. van Schaik, C. Jin, T. Hamilton, B. Torben-Nielsen and B. Webb Directional hearing in a silicon cricket. Biosystems, (in revision), 2005b [7] T. Delbrück and A. van Schaik, Bias Current Generators with Wide Dynamic Range, Analog Integrated Circuits and Signal Processing 42(2), 2005 [8] A. van Schaik and C. Jin, The Tau Cell: A New Method for the Implementation of Arbitrary Differential Equations, IEEE International Symposium on Circuits and Systems (ISCAS) 2003 [9] Kazuo Imaizumi and Gerald S. Pollack. Neural coding of sound frequency by cricket auditory receptors. The Journal of Neuroscience, 19(4):1508– 1516, 1999. [10] Berthold Hedwig and James F.A. Poulet. Complex auditory behaviour emerges from simple reactive steering. Nature, 430:781–785, 2004. [11] R. Reeve, B. Webb, A. Horchler, G. Indiveri, and R. Quinn. New technologies for testing a model of cricket phonotaxis on an outdoor robot platform. Robotics and Autonomous Systems, 51(1):41-54, 2005.
5 0.52426016 40 nips-2005-CMOL CrossNets: Possible Neuromorphic Nanoelectronic Circuits
Author: Jung Hoon Lee, Xiaolong Ma, Konstantin K. Likharev
Abstract: Hybrid “CMOL” integrated circuits, combining CMOS subsystem with nanowire crossbars and simple two-terminal nanodevices, promise to extend the exponential Moore-Law development of microelectronics into the sub-10-nm range. We are developing neuromorphic network (“CrossNet”) architectures for this future technology, in which neural cell bodies are implemented in CMOS, nanowires are used as axons and dendrites, while nanodevices (bistable latching switches) are used as elementary synapses. We have shown how CrossNets may be trained to perform pattern recovery and classification despite the limitations imposed by the CMOL hardware. Preliminary estimates have shown that CMOL CrossNets may be extremely dense (~10 7 cells per cm2) and operate approximately a million times faster than biological neural networks, at manageable power consumption. In Conclusion, we discuss in brief possible short-term and long-term applications of the emerging technology. 1 Introduction: CMOL Circuits Recent results [1, 2] indicate that the current VLSI paradigm based on CMOS technology can be hardly extended beyond the 10-nm frontier: in this range the sensitivity of parameters (most importantly, the gate voltage threshold) of silicon field-effect transistors to inevitable fabrication spreads grows exponentially. This sensitivity will probably send the fabrication facilities costs skyrocketing, and may lead to the end of Moore’s Law some time during the next decade. There is a growing consensus that the impending Moore’s Law crisis may be preempted by a radical paradigm shift from the purely CMOS technology to hybrid CMOS/nanodevice circuits, e.g., those of “CMOL” variety (Fig. 1). Such circuits (see, e.g., Ref. 3 for their recent review) would combine a level of advanced CMOS devices fabricated by the lithographic patterning, and two-layer nanowire crossbar formed, e.g., by nanoimprint, with nanowires connected by simple, similar, two-terminal nanodevices at each crosspoint. For such devices, molecular single-electron latching switches [4] are presently the leading candidates, in particular because they may be fabricated using the self-assembled monolayer (SAM) technique which already gave reproducible results for simpler molecular devices [5]. (a) nanodevices nanowiring and nanodevices interface pins upper wiring level of CMOS stack (b) βFCMOS Fnano α Fig. 1. CMOL circuit: (a) schematic side view, and (b) top-view zoom-in on several adjacent interface pins. (For clarity, only two adjacent nanodevices are shown.) In order to overcome the CMOS/nanodevice interface problems pertinent to earlier proposals of hybrid circuits [6], in CMOL the interface is provided by pins that are distributed all over the circuit area, on the top of the CMOS stack. This allows to use advanced techniques of nanowire patterning (like nanoimprint) which do not have nanoscale accuracy of layer alignment [3]. The vital feature of this interface is the tilt, by angle α = arcsin(Fnano/βFCMOS), of the nanowire crossbar relative to the square arrays of interface pins (Fig. 1b). Here Fnano is the nanowiring half-pitch, FCMOS is the half-pitch of the CMOS subsystem, and β is a dimensionless factor larger than 1 that depends on the CMOS cell complexity. Figure 1b shows that this tilt allows the CMOS subsystem to address each nanodevice even if Fnano << βFCMOS. By now, it has been shown that CMOL circuits can combine high performance with high defect tolerance (which is necessary for any circuit using nanodevices) for several digital applications. In particular, CMOL circuits with defect rates below a few percent would enable terabit-scale memories [7], while the performance of FPGA-like CMOL circuits may be several hundred times above that of overcome purely CMOL FPGA (implemented with the same FCMOS), at acceptable power dissipation and defect tolerance above 20% [8]. In addition, the very structure of CMOL circuits makes them uniquely suitable for the implementation of more complex, mixed-signal information processing systems, including ultradense and ultrafast neuromorphic networks. The objective of this paper is to describe in brief the current status of our work on the development of so-called Distributed Crossbar Networks (“CrossNets”) that could provide high performance despite the limitations imposed by CMOL hardware. A more detailed description of our earlier results may be found in Ref. 9. 2 Synapses The central device of CrossNet is a two-terminal latching switch [3, 4] (Fig. 2a) which is a combination of two single-electron devices, a transistor and a trap [3]. The device may be naturally implemented as a single organic molecule (Fig. 2b). Qualitatively, the device operates as follows: if voltage V = Vj – Vk applied between the external electrodes (in CMOL, nanowires) is low, the trap island has no net electric charge, and the single-electron transistor is closed. If voltage V approaches certain threshold value V+ > 0, an additional electron is inserted into the trap island, and its field lifts the Coulomb blockade of the single-electron transistor, thus connecting the nanowires. The switch state may be reset (e.g., wires disconnected) by applying a lower voltage V < V- < V+. Due to the random character of single-electron tunneling [2], the quantitative description of the switch is by necessity probabilistic: actually, V determines only the rates Γ↑↓ of device switching between its ON and OFF states. The rates, in turn, determine the dynamics of probability p to have the transistor opened (i.e. wires connected): dp/dt = Γ↑(1 - p) - Γ↓p. (1) The theory of single-electron tunneling [2] shows that, in a good approximation, the rates may be presented as Γ↑↓ = Γ0 exp{±e(V - S)/kBT} , (2) (a) single-electron trap tunnel junction Vj Vk single-electron transistor (b) O clipping group O N C R diimide acceptor groups O O C N R R O OPE wires O N R R N O O R O N R R = hexyl N O O R R O N C R R R Fig. 2. (a) Schematics and (b) possible molecular implementation of the two-terminal single-electron latching switch where Γ0 and S are constants depending on physical parameters of the latching switches. Note that despite the random character of switching, the strong nonlinearity of Eq. (2) allows to limit the degree of the device “fuzziness”. 3 CrossNets Figure 3a shows the generic structure of a CrossNet. CMOS-implemented somatic cells (within the Fire Rate model, just nonlinear differential amplifiers, see Fig. 3b,c) apply their output voltages to “axonic” nanowires. If the latching switch, working as an elementary synapse, on the crosspoint of an axonic wire with the perpendicular “dendritic” wire is open, some current flows into the latter wire, charging it. Since such currents are injected into each dendritic wire through several (many) open synapses, their addition provides a natural passive analog summation of signals from the corresponding somas, typical for all neural networks. Examining Fig. 3a, please note the open-circuit terminations of axonic and dendritic lines at the borders of the somatic cells; due to these terminations the somas do not communicate directly (but only via synapses). The network shown on Fig. 3 is evidently feedforward; recurrent networks are achieved in the evident way by doubling the number of synapses and nanowires per somatic cell (Fig. 3c). Moreover, using dual-rail (bipolar) representation of the signal, and hence doubling the number of nanowires and elementary synapses once again, one gets a CrossNet with somas coupled by compact 4-switch groups [9]. Using Eqs. (1) and (2), it is straightforward to show that that the average synaptic weight wjk of the group obeys the “quasi-Hebbian” rule: d w jk = −4Γ0 sinh (γ S ) sinh (γ V j ) sinh (γ Vk ) . dt (3) (a) - +soma j (b) RL + -- jk+ RL (c) jk- RL + -- -+soma k RL Fig. 3. (a) Generic structure of the simplest, (feedforward, non-Hebbian) CrossNet. Red lines show “axonic”, and blue lines “dendritic” nanowires. Gray squares are interfaces between nanowires and CMOS-based somas (b, c). Signs show the dendrite input polarities. Green circles denote molecular latching switches forming elementary synapses. Bold red and blue points are open-circuit terminations of the nanowires, that do not allow somas to interact in bypass of synapses In the simplest cases (e.g., quasi-Hopfield networks with finite connectivity), the tri-level synaptic weights of the generic CrossNets are quite satisfactory, leading to just a very modest (~30%) network capacity loss. However, some applications (in particular, pattern classification) may require a larger number of weight quantization levels L (e.g., L ≈ 30 for a 1% fidelity [9]). This may be achieved by using compact square arrays (e.g., 4×4) of latching switches (Fig. 4). Various species of CrossNets [9] differ also by the way the somatic cells are distributed around the synaptic field. Figure 5 shows feedforward versions of two CrossNet types most explored so far: the so-called FlossBar and InBar. The former network is more natural for the implementation of multilayered perceptrons (MLP), while the latter system is preferable for recurrent network implementations and also allows a simpler CMOS design of somatic cells. The most important advantage of CrossNets over the hardware neural networks suggested earlier is that these networks allow to achieve enormous density combined with large cell connectivity M >> 1 in quasi-2D electronic circuits. 4 CrossNet training CrossNet training faces several hardware-imposed challenges: (i) The synaptic weight contribution provided by the elementary latching switch is binary, so that for most applications the multi-switch synapses (Fig. 4) are necessary. (ii) The only way to adjust any particular synaptic weight is to turn ON or OFF the corresponding latching switch(es). This is only possible to do by applying certain voltage V = Vj – Vk between the two corresponding nanowires. At this procedure, other nanodevices attached to the same wires should not be disturbed. (iii) As stated above, synapse state switching is a statistical progress, so that the degree of its “fuzziness” should be carefully controlled. (a) Vj (b) V w – A/2 i=1 i=1 2 2 … … n n Vj V w+ A/2 i' = 1 RL 2 … i' = 1 n RS ±(V t –A/2) 2 … RS n ±(V t +A/2) Fig. 4. Composite synapse for providing L = 2n2+1 discrete levels of the weight in (a) operation and (b) weight adjustment modes. The dark-gray rectangles are resistive metallic strips at soma/nanowire interfaces (a) (b) Fig. 5. Two main CrossNet species: (a) FlossBar and (b) InBar, in the generic (feedforward, non-Hebbian, ternary-weight) case for the connectivity parameter M = 9. Only the nanowires and nanodevices coupling one cell (indicated with red dashed lines) to M post-synaptic cells (blue dashed lines) are shown; actually all the cells are similarly coupled We have shown that these challenges may be met using (at least) the following training methods [9]: (i) Synaptic weight import. This procedure is started with training of a homomorphic “precursor” artificial neural network with continuous synaptic weighs wjk, implemented in software, using one of established methods (e.g., error backpropagation). Then the synaptic weights wjk are transferred to the CrossNet, with some “clipping” (rounding) due to the binary nature of elementary synaptic weights. To accomplish the transfer, pairs of somatic cells are sequentially selected via CMOS-level wiring. Using the flexibility of CMOS circuitry, these cells are reconfigured to apply external voltages ±VW to the axonic and dendritic nanowires leading to a particular synapse, while all other nanowires are grounded. The voltage level V W is selected so that it does not switch the synapses attached to only one of the selected nanowires, while voltage 2VW applied to the synapse at the crosspoint of the selected wires is sufficient for its reliable switching. (In the composite synapses with quasi-continuous weights (Fig. 4), only a part of the corresponding switches is turned ON or OFF.) (ii) Error backpropagation. The synaptic weight import procedure is straightforward when wjk may be simply calculated, e.g., for the Hopfield-type networks. However, for very large CrossNets used, e.g., as pattern classifiers the precursor network training may take an impracticably long time. In this case the direct training of a CrossNet may become necessary. We have developed two methods of such training, both based on “Hebbian” synapses consisting of 4 elementary synapses (latching switches) whose average weight dynamics obeys Eq. (3). This quasi-Hebbian rule may be used to implement the backpropagation algorithm either using a periodic time-multiplexing [9] or in a continuous fashion, using the simultaneous propagation of signals and errors along the same dual-rail channels. As a result, presently we may state that CrossNets may be taught to perform virtually all major functions demonstrated earlier with the usual neural networks, including the corrupted pattern restoration in the recurrent quasi-Hopfield mode and pattern classification in the feedforward MLP mode [11]. 5 C r o s s N e t p e r f o r m an c e e s t i m a t e s The significance of this result may be only appreciated in the context of unparalleled physical parameters of CMOL CrossNets. The only fundamental limitation on the half-pitch Fnano (Fig. 1) comes from quantum-mechanical tunneling between nanowires. If the wires are separated by vacuum, the corresponding specific leakage conductance becomes uncomfortably large (~10-12 Ω-1m-1) only at Fnano = 1.5 nm; however, since realistic insulation materials (SiO2, etc.) provide somewhat lower tunnel barriers, let us use a more conservative value Fnano= 3 nm. Note that this value corresponds to 1012 elementary synapses per cm2, so that for 4M = 104 and n = 4 the areal density of neural cells is close to 2×107 cm-2. Both numbers are higher than those for the human cerebral cortex, despite the fact that the quasi-2D CMOL circuits have to compete with quasi-3D cerebral cortex. With the typical specific capacitance of 3×10-10 F/m = 0.3 aF/nm, this gives nanowire capacitance C0 ≈ 1 aF per working elementary synapse, because the corresponding segment has length 4Fnano. The CrossNet operation speed is determined mostly by the time constant τ0 of dendrite nanowire capacitance recharging through resistances of open nanodevices. Since both the relevant conductance and capacitance increase similarly with M and n, τ0 ≈ R0C0. The possibilities of reduction of R0, and hence τ0, are limited mostly by acceptable power dissipation per unit area, that is close to Vs2/(2Fnano)2R0. For room-temperature operation, the voltage scale V0 ≈ Vt should be of the order of at least 30 kBT/e ≈ 1 V to avoid thermally-induced errors [9]. With our number for Fnano, and a relatively high but acceptable power consumption of 100 W/cm2, we get R0 ≈ 1010Ω (which is a very realistic value for single-molecule single-electron devices like one shown in Fig. 3). With this number, τ0 is as small as ~10 ns. This means that the CrossNet speed may be approximately six orders of magnitude (!) higher than that of the biological neural networks. Even scaling R0 up by a factor of 100 to bring power consumption to a more comfortable level of 1 W/cm2, would still leave us at least a four-orders-of-magnitude speed advantage. 6 D i s c u s s i on: P o s s i bl e a p p l i c at i o n s These estimates make us believe that that CMOL CrossNet chips may revolutionize the neuromorphic network applications. Let us start with the example of relatively small (1-cm2-scale) chips used for recognition of a face in a crowd [11]. The most difficult feature of such recognition is the search for face location, i.e. optimal placement of a face on the image relative to the panel providing input for the processing network. The enormous density and speed of CMOL hardware gives a possibility to time-and-space multiplex this task (Fig. 6). In this approach, the full image (say, formed by CMOS photodetectors on the same chip) is divided into P rectangular panels of h×w pixels, corresponding to the expected size and approximate shape of a single face. A CMOS-implemented communication channel passes input data from each panel to the corresponding CMOL neural network, providing its shift in time, say using the TV scanning pattern (red line in Fig. 6). The standard methods of image classification require the network to have just a few hidden layers, so that the time interval Δt necessary for each mapping position may be so short that the total pattern recognition time T = hwΔt may be acceptable even for online face recognition. w h image network input Fig. 6. Scan mapping of the input image on CMOL CrossNet inputs. Red lines show the possible time sequence of image pixels sent to a certain input of the network processing image from the upper-left panel of the pattern Indeed, let us consider a 4-Megapixel image partitioned into 4K 32×32-pixel panels (h = w = 32). This panel will require an MLP net with several (say, four) layers with 1K cells each in order to compare the panel image with ~10 3 stored faces. With the feasible 4-nm nanowire half-pitch, and 65-level synapses (sufficient for better than 99% fidelity [9]), each interlayer crossbar would require chip area about (4K×64 nm)2 = 64×64 μm2, fitting 4×4K of them on a ~0.6 cm2 chip. (The CMOS somatic-layer and communication-system overheads are negligible.) With the acceptable power consumption of the order of 10 W/cm2, the input-to-output signal propagation in such a network will take only about 50 ns, so that Δt may be of the order of 100 ns and the total time T = hwΔt of processing one frame of the order of 100 microseconds, much shorter than the typical TV frame time of ~10 milliseconds. The remaining two-orders-of-magnitude time gap may be used, for example, for double-checking the results via stopping the scan mapping (Fig. 6) at the most promising position. (For this, a simple feedback from the recognition output to the mapping communication system is necessary.) It is instructive to compare the estimated CMOL chip speed with that of the implementation of a similar parallel network ensemble on a CMOS signal processor (say, also combined on the same chip with an array of CMOS photodetectors). Even assuming an extremely high performance of 30 billion additions/multiplications per second, we would need ~4×4K×1K×(4K)2/(30×109) ≈ 104 seconds ~ 3 hours per frame, evidently incompatible with the online image stream processing. Let us finish with a brief (and much more speculative) discussion of possible long-term prospects of CMOL CrossNets. Eventually, large-scale (~30×30 cm2) CMOL circuits may become available. According to the estimates given in the previous section, the integration scale of such a system (in terms of both neural cells and synapses) will be comparable with that of the human cerebral cortex. Equipped with a set of broadband sensor/actuator interfaces, such (necessarily, hierarchical) system may be capable, after a period of initial supervised training, of further self-training in the process of interaction with environment, with the speed several orders of magnitude higher than that of its biological prototypes. Needless to say, the successful development of such self-developing systems would have a major impact not only on all information technologies, but also on the society as a whole. Acknowledgments This work has been supported in part by the AFOSR, MARCO (via FENA Center), and NSF. Valuable contributions made by Simon Fölling, Özgür Türel and Ibrahim Muckra, as well as useful discussions with P. Adams, J. Barhen, D. Hammerstrom, V. Protopopescu, T. Sejnowski, and D. Strukov are gratefully acknowledged. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] Frank, D. J. et al. (2001) Device scaling limits of Si MOSFETs and their application dependencies. Proc. IEEE 89(3): 259-288. Likharev, K. K. (2003) Electronics below 10 nm, in J. Greer et al. (eds.), Nano and Giga Challenges in Microelectronics, pp. 27-68. Amsterdam: Elsevier. Likharev, K. K. and Strukov, D. B. (2005) CMOL: Devices, circuits, and architectures, in G. Cuniberti et al. (eds.), Introducing Molecular Electronics, Ch. 16. Springer, Berlin. Fölling, S., Türel, Ö. & Likharev, K. K. (2001) Single-electron latching switches as nanoscale synapses, in Proc. of the 2001 Int. Joint Conf. on Neural Networks, pp. 216-221. Mount Royal, NJ: Int. Neural Network Society. Wang, W. et al. (2003) Mechanism of electron conduction in self-assembled alkanethiol monolayer devices. Phys. Rev. B 68(3): 035416 1-8. Stan M. et al. (2003) Molecular electronics: From devices and interconnect to circuits and architecture, Proc. IEEE 91(11): 1940-1957. Strukov, D. B. & Likharev, K. K. (2005) Prospects for terabit-scale nanoelectronic memories. Nanotechnology 16(1): 137-148. Strukov, D. B. & Likharev, K. K. (2005) CMOL FPGA: A reconfigurable architecture for hybrid digital circuits with two-terminal nanodevices. Nanotechnology 16(6): 888-900. Türel, Ö. et al. (2004) Neuromorphic architectures for nanoelectronic circuits”, Int. J. of Circuit Theory and Appl. 32(5): 277-302. See, e.g., Hertz J. et al. (1991) Introduction to the Theory of Neural Computation. Cambridge, MA: Perseus. Lee, J. H. & Likharev, K. K. (2005) CrossNets as pattern classifiers. Lecture Notes in Computer Sciences 3575: 434-441.
6 0.50597841 118 nips-2005-Learning in Silicon: Timing is Everything
8 0.37437621 169 nips-2005-Saliency Based on Information Maximization
9 0.37367475 88 nips-2005-Gradient Flow Independent Component Analysis in Micropower VLSI
10 0.37334949 17 nips-2005-Active Bidirectional Coupling in a Cochlear Chip
11 0.37325561 8 nips-2005-A Criterion for the Convergence of Learning with Spike Timing Dependent Plasticity
12 0.37120092 157 nips-2005-Principles of real-time computing with feedback applied to cortical microcircuit models
13 0.36471289 181 nips-2005-Spiking Inputs to a Winner-take-all Network
14 0.3519097 188 nips-2005-Temporally changing synaptic plasticity
15 0.34145552 149 nips-2005-Optimal cue selection strategy
16 0.34128419 194 nips-2005-Top-Down Control of Visual Attention: A Rational Account
17 0.33846676 128 nips-2005-Modeling Memory Transfer and Saving in Cerebellar Motor Learning
18 0.33268556 143 nips-2005-Off-Road Obstacle Avoidance through End-to-End Learning
19 0.3252869 203 nips-2005-Visual Encoding with Jittering Eyes
20 0.29213151 193 nips-2005-The Role of Top-down and Bottom-up Processes in Guiding Eye Movements during Visual Search
topicId topicWeight
[(3, 0.025), (10, 0.027), (27, 0.027), (31, 0.043), (34, 0.05), (39, 0.036), (42, 0.365), (55, 0.019), (57, 0.045), (60, 0.016), (65, 0.011), (69, 0.059), (73, 0.019), (77, 0.016), (88, 0.033), (91, 0.049), (99, 0.033)]
simIndex simValue paperId paperTitle
same-paper 1 0.89365834 176 nips-2005-Silicon growth cones map silicon retina
Author: Brian Taba, Kwabena Boahen
Abstract: We demonstrate the first fully hardware implementation of retinotopic self-organization, from photon transduction to neural map formation. A silicon retina transduces patterned illumination into correlated spike trains that drive a population of silicon growth cones to automatically wire a topographic mapping by migrating toward sources of a diffusible guidance cue that is released by postsynaptic spikes. We varied the pattern of illumination to steer growth cones projected by different retinal ganglion cell types to self-organize segregated or coordinated retinotopic maps. 1
2 0.76709253 68 nips-2005-Factorial Switching Kalman Filters for Condition Monitoring in Neonatal Intensive Care
Author: Christopher Williams, John Quinn, Neil Mcintosh
Abstract: The observed physiological dynamics of an infant receiving intensive care are affected by many possible factors, including interventions to the baby, the operation of the monitoring equipment and the state of health. The Factorial Switching Kalman Filter can be used to infer the presence of such factors from a sequence of observations, and to estimate the true values where these observations have been corrupted. We apply this model to clinical time series data and show it to be effective in identifying a number of artifactual and physiological patterns. 1
3 0.31909859 181 nips-2005-Spiking Inputs to a Winner-take-all Network
Author: Matthias Oster, Shih-Chii Liu
Abstract: Recurrent networks that perform a winner-take-all computation have been studied extensively. Although some of these studies include spiking networks, they consider only analog input rates. We present results of this winner-take-all computation on a network of integrate-and-fire neurons which receives spike trains as inputs. We show how we can configure the connectivity in the network so that the winner is selected after a pre-determined number of input spikes. We discuss spiking inputs with both regular frequencies and Poisson-distributed rates. The robustness of the computation was tested by implementing the winner-take-all network on an analog VLSI array of 64 integrate-and-fire neurons which have an innate variance in their operating parameters. 1
4 0.29517692 11 nips-2005-A Hierarchical Compositional System for Rapid Object Detection
Author: Long Zhu, Alan L. Yuille
Abstract: We describe a hierarchical compositional system for detecting deformable objects in images. Objects are represented by graphical models. The algorithm uses a hierarchical tree where the root of the tree corresponds to the full object and lower-level elements of the tree correspond to simpler features. The algorithm proceeds by passing simple messages up and down the tree. The method works rapidly, in under a second, on 320 × 240 images. We demonstrate the approach on detecting cats, horses, and hands. The method works in the presence of background clutter and occlusions. Our approach is contrasted with more traditional methods such as dynamic programming and belief propagation. 1
5 0.29471314 96 nips-2005-Inference with Minimal Communication: a Decision-Theoretic Variational Approach
Author: O. P. Kreidl, Alan S. Willsky
Abstract: Given a directed graphical model with binary-valued hidden nodes and real-valued noisy observations, consider deciding upon the maximum a-posteriori (MAP) or the maximum posterior-marginal (MPM) assignment under the restriction that each node broadcasts only to its children exactly one single-bit message. We present a variational formulation, viewing the processing rules local to all nodes as degrees-of-freedom, that minimizes the loss in expected (MAP or MPM) performance subject to such online communication constraints. The approach leads to a novel message-passing algorithm to be executed offline, or before observations are realized, which mitigates the performance loss by iteratively coupling all rules in a manner implicitly driven by global statistics. We also provide (i) illustrative examples, (ii) assumptions that guarantee convergence and efficiency and (iii) connections to active research areas. 1
6 0.29235959 8 nips-2005-A Criterion for the Convergence of Learning with Spike Timing Dependent Plasticity
7 0.28783587 106 nips-2005-Large-scale biophysical parameter estimation in single neurons via constrained linear regression
8 0.28480676 157 nips-2005-Principles of real-time computing with feedback applied to cortical microcircuit models
9 0.28478876 90 nips-2005-Hot Coupling: A Particle Approach to Inference and Normalization on Pairwise Undirected Graphs
10 0.2838082 67 nips-2005-Extracting Dynamical Structure Embedded in Neural Activity
11 0.2801072 61 nips-2005-Dynamical Synapses Give Rise to a Power-Law Distribution of Neuronal Avalanches
12 0.27787167 149 nips-2005-Optimal cue selection strategy
13 0.27713504 124 nips-2005-Measuring Shared Information and Coordinated Activity in Neuronal Networks
14 0.27697542 200 nips-2005-Variable KD-Tree Algorithms for Spatial Pattern Search and Discovery
15 0.27588955 1 nips-2005-AER Building Blocks for Multi-Layer Multi-Chip Neuromorphic Vision Systems
16 0.27584761 72 nips-2005-Fast Online Policy Gradient Learning with SMD Gain Vector Adaptation
17 0.27462512 30 nips-2005-Assessing Approximations for Gaussian Process Classification
18 0.27334991 43 nips-2005-Comparing the Effects of Different Weight Distributions on Finding Sparse Representations
19 0.27131888 78 nips-2005-From Weighted Classification to Policy Search
20 0.27119231 99 nips-2005-Integrate-and-Fire models with adaptation are good enough