nips nips2002 nips2002-160 knowledge-graph by maker-knowledge-mining

160 nips-2002-Optoelectronic Implementation of a FitzHugh-Nagumo Neural Model

Source: pdf

Author: Alexandre R. Romariz, Kelvin Wagner

Abstract: An optoelectronic implementation of a spiking neuron model based on the FitzHugh-Nagumo equations is presented. A tunable semiconductor laser source and a spectral ﬁlter provide a nonlinear mapping from driver voltage to detected signal. Linear electronic feedback completes the implementation, which allows either electronic or optical input signals. Experimental results for a single system and numeric results of model interaction conﬁrm that important features of spiking neural models can be implemented through this approach.

Reference: text

Summary: the most important sentenses genereted by tfidf model

sentIndex sentText sentNum sentScore

1 edu Abstract An optoelectronic implementation of a spiking neuron model based on the FitzHugh-Nagumo equations is presented. [sent-4, score-0.531]

2 A tunable semiconductor laser source and a spectral ﬁlter provide a nonlinear mapping from driver voltage to detected signal. [sent-5, score-0.897]

3 Linear electronic feedback completes the implementation, which allows either electronic or optical input signals. [sent-6, score-0.959]

4 Experimental results for a single system and numeric results of model interaction conﬁrm that important features of spiking neural models can be implemented through this approach. [sent-7, score-0.209]

5 A computational paradigm that takes into account the timing of spikes (instead of spike rates only) might be more efﬁcient for signal representation and processing, especially at short time windows [3, 4, 5]. [sent-10, score-0.235]

6 However, the implementation of digital primitives have not as yet proved competitive against the scalability and low power operation of digital electronic gates. [sent-12, score-0.341]

7 It is then natural to explore the features of optics for different computational paradigms. [sent-13, score-0.196]

8 Optical implementations of Artiﬁcial Neural Networks have to deal with the problem of representing the nonlinear activation functions that deﬁne the input-output mappings for each neuron. [sent-15, score-0.122]

9 Although nonlinear optics has been suggested for implementing neurons, hybrid optoelectronic systems, where the task of producing nonlinearity is given to the electronic circuits, may be more practical [10, 11]. [sent-16, score-0.756]

10 In the case of pulsing neurons, the task seems more difﬁcult still, for instead of a nonlinear static map we are required to implement a nonlinear dynamical system. [sent-17, score-0.386]

11 In this paper we demonstrate and evaluate an optoelectronic implementation of an artiﬁcial spiking neuron, based on the FitzHugh-Nagumo equations. [sent-19, score-0.462]

12 The proposed implementation uses wavelength tunability of a laser source and a birefringent crystal to produce a nonlinear mapping from driving voltage to detected optical output [15]. [sent-20, score-1.611]

13 Linear electronic feedback to the laser drive current completes the physical implementation of this model neuron. [sent-21, score-0.569]

14 Inputs can be presented optically or electronically, and output signals are also readily available as optical or electronic pulses. [sent-22, score-0.658]

15 Section 2 reviews the FitzHugh-Nagumo equations and describes the particular optoelectronic spiking neuron implementation we propose here. [sent-24, score-0.531]

16 In Section 3 we analyze and illustrate dynamical properties of the model. [sent-25, score-0.093]

17 Experimental results of the optoelectronic system implementing one model are presented in Section 4. [sent-26, score-0.325]

18 2 Modiﬁed FN Neural Model and optoelectronic implementation The FitzHugh-Nagumo neuron model [16, 17] is appealing for physical implementation, as it is fairly simple and completely described by a pair of coupled differential equations: ¨¦ % # ¨¦ ©'&$©"! [sent-28, score-0.454]

19 §©§¥ ¨¦ 7 # ¨¦ 865"4¤©§32 ¨¦ ¥ (1) £ ©§¥ ¢ ¨ ¦ ¤¡ £ ©1 ) ¨ ¦ 0( ¥ where is an excitable state variable that exhibits bi-stability as a result of the nonlinear term, and is a linear recovery variable, bringing the neuron back to a resting state. [sent-29, score-0.265]

20 This model has been previously implemented in CMOS integrated electronics [18]. [sent-31, score-0.089]

21 @¥ 9¥ In optical implementation of neural networks, the required nonlinear functions are usually performed through electronic devices, with adaptive linear interconnection done in the optical domain. [sent-32, score-1.392]

22 We here explore the possibility of optical implementation of the required by using the nonlinear response of linear optical systems to varianonlinear function tions of the wavelength. [sent-33, score-1.173]

23 BA Consider a birefringent material placed between crossed polarizers. [sent-34, score-0.163]

24 p V W5¦ r V V¦ W5US In semiconductor lasers, and Vertical Cavity Surface Emitting Lasers (VCSELs) in particular, an input current produces a small modulation in the radiation wavelength . [sent-36, score-0.444]

25 Linearizing the variation in Equation 2, we ﬁnd a nonlinear mapping from driving voltage to detected signal: V u©¦ r (3) V ¥ Y§15¦ s e d b` X ¥¦ ¥¦ FD h f e©UyY5x EC ¥ r w tv w Detected Optical Signal 0. [sent-37, score-0.455]

26 35 (b) Figure 1: a Experimental setup for the wavelength-based nonlinear oscillator, with simpliﬁed view of the electronic feedback. [sent-50, score-0.238]

27 b Experimental evidence of nonlinear mapping from driver voltage to detected signal (open loop), as a result of wavelength modulation as well as laser threshold and saturation. [sent-51, score-1.112]

28 V ¥ where is the driving voltage (linearly converted to an input current through the driver includes all conversion factors in the detection transconductance) and the function process, as well as nonlinear phenomena such as laser threshold and saturation. [sent-52, score-0.746]

29 ¥¦ ©U A simple nonlinear feedback loop can now be established, by feeding the detected signal back to the driver. [sent-53, score-0.341]

30 This basic arrangement has been used to investigate chaotic behavior in delayed-feedback tunable lasers [15] . [sent-54, score-0.228]

31 It is used here as the nonlinearity for an optical self-pulsing mechanism in order to implement neural-like pulses based on the following dynamical system ( ) ¨¦ % # ¨¦ ©'&6©"! [sent-55, score-0.708]

32 §©§¥ ¨ ¥ ¨¦ 7 # ¨¦ &'5"4¤©§32 ¨¦ ¥ (4) £ ¨ ¦ ¤¡ ©§¥ ¢ ¨¦ £ ©" ( ¥ ¡ t Again is a fast state variable, and a relatively slow recovery variable, so that . [sent-57, score-0.082]

33 Light from the tunable source is collimated and propagates through a piece of birefringent crystal. [sent-59, score-0.192]

34 The crystal fast and slow axis are at 45 degrees to the polarizer and analyzer passing axis. [sent-60, score-0.123]

35 The effective propagation length through the crystal (and corresponding wavelength selectivity) is doubled with the use of a mirror. [sent-61, score-0.293]

36 A simpliﬁed view of the electronic feedback is also shown. [sent-63, score-0.253]

37 Leaky integrators and linear analog summations implement the linear part of Equation 4, while the nonlinear response (in intensity) of the . [sent-64, score-0.235]

38 optical ﬁlter implements ¡ ¥¦ Y5 A VCSEL was used as tunable laser source. [sent-65, score-0.727]

39 These vertical-cavity semiconductor lasers have, when compared to edge-emitting diode lasers, larger separation between longitudinal modes, more circularly-symmetric beams and lower fabrication costs [19]. [sent-66, score-0.231]

40 As the input current is increased, the heating of the cavity red-shifts the resonant wavelength [20], and this is the main mechanism we are exploring for wavelength modulation. [sent-67, score-0.6]

41 An experimental veriﬁcation of the expected sinusoidal variation of detected power with modulation voltage is given in Figure 1b. [sent-68, score-0.571]

42 A slow (800Hz) modulation ramp was applied to the driver, and the detected power variation was acquired. [sent-69, score-0.316]

43 Unlike the experiment with a DBR laser diode reported by Goedgebuer et al. [sent-71, score-0.23]

44 [15], it is apparent that current modulation is affecting not only wavelength (and hence effective optical path difference among polarization components) but overall output power as well. [sent-72, score-0.91]

45 Modulation depth is limited (non-zero troughs in the sinusoidal variation), which we attribute to the multiple transverse modes that the device supports. [sent-73, score-0.067]

46 However, as we are going to be operating near Figure 2: Continuous line: trajectory of the system under strong input, obtained by numeric integration ( -order Runge-Kutta) of Equation 4. [sent-74, score-0.158]

47 Stability analysis show that the equilibrium point where the dotted line: nullcline nullclines meet is unstable, so the limit cycle is the sole attractor. [sent-78, score-0.115]

48 ¡ ¢F £ ¤&¥ £ ¡ ) £ v £ ¥ ( ¨ v c¦ £ g§ §£ ¦ £ ¨ ¨ §©¦ £ £ % ¨¦ ©§v 7 £ v H2 the ﬁrst maximum (see Section 3), the power variation over successive maxima should not affect the dynamical properties of the closed-loop system. [sent-80, score-0.182]

49 The relatively smooth curve obtained indicates that no mode hops occurred for this driving current range, which was indeed conﬁrmed with Optical Spectrum Analyzer measurements. [sent-81, score-0.062]

50 3 Simulations FitzHugh-Nagumo models are known to have so-called class II neural excitability (see [21] for a review). [sent-82, score-0.054]

51 This class is characterized by an Andronov-Hopf bifurcation for increasing excitation, and exhibits some dynamical phenomena that are not present in integrate-andﬁre dynamics. [sent-83, score-0.186]

52 For equal intensity input pulses, integrators will respond maximally to the pulse train with lowest inter-spike interval. [sent-84, score-0.371]

53 Class II neurons have resonant response to a range of input frequencies. [sent-85, score-0.182]

54 There are non-trivial forms of excitation in resonator models that are not matched by integrators: the former can produce a spike at the end of an inhibitory pulse, and conversely, can have a limit cycle condition interrupted (with the system recovering to rest) by an excitatory pulse. [sent-86, score-0.265]

55 We have veriﬁed that these characteristics are maintained in the modiﬁed optical model, despite the use of a sinusoidal nonlinearity instead of the original degree polynomial function. [sent-87, score-0.569]

56 Stability analysis based on the Jacobian of the dynamical system (Equation 4) shows an Andronov-Hopf bifurcation, as in the original model. [sent-88, score-0.13]

57 Limit cycle interruption through exciting pulses is shown in Section 5. [sent-89, score-0.079]

58 Parameters were chosen so that a typical excursion in modulation voltage goes from the dead zone (below the lasing threshold) to around the ﬁrst peak in the nonlinear detector transfer function. [sent-91, score-0.435]

59 This is an interesting choice because the optical output is only present during spiking, and can be used directly as an input to other optoelectronic neurons. [sent-92, score-0.805]

60 000 10 20 30 (a) (b) Figure 3: Dynamical system response to strong constant input. [sent-108, score-0.085]

61 The double-peak in the optical variable can be understood by following the trajectory indicated in Figure 2, bearing in mind the non-monotonic mapping from driver voltage to detected signal. [sent-130, score-1.012]

62 The decrease in driver voltage observed as the recovery variable increases produces initially an increase in detected power, and thus the second, broader peak at the end of the cycle. [sent-131, score-0.568]

63 The production of sustained oscillations for constant input is one of the desired characteristics of the model, but in a network, neurons will mostly communicate through their pulsed output. [sent-132, score-0.199]

64 The response of the system to pulsed inputs can be seen in Figure 4. [sent-133, score-0.195]

65 The output optical signal response is all-or-none, but sub-threshold integration of weak inputs is being performed, as the waveform for driver voltage shows in the ﬁrst pulse. [sent-134, score-0.958]

66 As slowly returns to 0, a new excitation just after a pulse is less likely, which can be seen at the response to the third pulse. [sent-135, score-0.343]

67 The experimentally observed waveforms agree with the simulations, though details of the pulsing in the optical output are different. [sent-136, score-0.635]

68 Pulse advance vs input pulse phase 4 ∆φo rad 2 Bias φi 0 No spikes 0 -2 2π -4 a 0 2 4 Input Pulse Phase (rad) 6 b Figure 5: Numeric illustration of the effect of input timing on the advance of the next spike, in the modiﬁed FitzHugh-Nagumo system. [sent-137, score-0.698]

69 The most elegant optical implementation of adaptive interconnection is through dynamic volume holography[6, 11], but that requires a set of coherent optical signals, not what we have with an array of pulse emitters. [sent-146, score-1.404]

70 In contrast, the matrix-vector multiplier architecture allows parallel interconnection of incoherent optical signals, and has been used to demonstrate implementations of the Hopﬁeld model [7] and Boltzman machines [9]. [sent-147, score-0.619]

71 An interesting aspect of the coupled dynamics in oscillators exhibiting class II excitability is that the timing of an input pulse can result in advance or retardation of the next spike [22]. [sent-148, score-0.752]

72 This is potentially relevant for hardware implementation, as the excitatory (i. [sent-149, score-0.076]

73 , inducing an early spike) or inhibitory character of the connection might be controlled without changing signs of the coupling strength. [sent-151, score-0.094]

74 In Figure 5 we show a simulation illustrating the effect of input pulse timing in advancing the output spike. [sent-152, score-0.418]

75 A constant input to a model neuron (Equation 4) was maintained, producing periodic spiking. [sent-153, score-0.163]

76 A second, positive, pulsed input was activated in between spikes, and the effect of this coupling on the advance or retardation of the next spike was veriﬁed ) with as the timing of the input was varied. [sent-154, score-0.583]

77 A region of output spike retardation ( excitatory pulsed input can be seen. [sent-155, score-0.412]

78 Even more interesting, for phases around rad relative to the latest spike, the excitatory pulse can terminate periodic spiking altogether. [sent-156, score-0.583]

79 For this particular condition, the equilibrium point of the system is stable. [sent-158, score-0.069]

80 When correctly timed, the short excitatory pulse forces the system out of its limit cycle, into the basin of attraction of the stable equilibrium, hence stopping the periodic spiking. [sent-159, score-0.404]

81 As the individual models used in this simulations were shown to match experimental implementations in Section 4, we expect to observe the same kind of effect in the coupling of the optoelectronic oscillators. [sent-160, score-0.417]

82 1600 140 a b % Figure 6: (a): Simulated response illustrating return to stability with excitatory pulse. [sent-190, score-0.156]

83 v v ¡9¦ £ 6 Ongoing work and conclusions Implementation of a modiﬁed FN neuron model with a nonlinear transfer function realized with a wavelength-tuned VCSEL source, a linear optical spectral ﬁlter and linear electronic feedback was demonstrated. [sent-196, score-0.906]

84 The system dynamical behavior agrees with simulated responses, and exhibits some of the basic features of neuron dynamics that are currently being investigated in the area of spiking neural networks. [sent-197, score-0.377]

85 Further experiments are being done to demonstrate coupling effects like the ones described in Section 5. [sent-198, score-0.094]

86 In particular, the use of external optical signals directly onto the detector to implement optical coupling has been demonstrated. [sent-199, score-1.125]

87 Feedback circuit simpliﬁcation is another important aspect, since we are interested in implementing large arrays of spiking neurons. [sent-200, score-0.149]

88 Results reported here were obtained at low frequency (1-100 KHz), limited by ampliﬁer and detector bandwidths. [sent-202, score-0.06]

89 With faster electronics and detectors, the limiting factor in this arrangement would be the time constant for thermal expansion of the VCSEL cavity, which is around 1 . [sent-203, score-0.156]

90 £ ¤¢ Even faster operation is possible when using the internal dynamics of wavelength modulation itself, instead of external electronic feedback. [sent-206, score-0.509]

91 In addition to the thermally-induced modulation of wavelength, carrier injection modiﬁes the index of refraction of the active region directly, which results in an opposite wavelength shift. [sent-207, score-0.389]

92 By using this carrier injection effect to implement the recovery variable, feedback electronics is simpliﬁed and a much faster time constant controls the model dynamics. [sent-208, score-0.371]

93 Optical coupling of VCSELs has the potential to generate over 40GHz pulsations [23]. [sent-209, score-0.094]

94 Our goal is to investigate those optical oscillators as a technology for implementing fast networks of spiking artiﬁcial neurons. [sent-210, score-0.707]

95 Lower bounds for the computational power of spiking neurons. [sent-231, score-0.159]

96 Pattern recognition computation using action potential timing for stimulus representation. [sent-236, score-0.079]

97 Implementation of a large-scale optical neural network by use of a coaxial lenslet array for interconnection. [sent-263, score-0.468]

98 Self-pulsing and chaos in an extended-cavity diode laser with intracavity atomic absorber. [sent-297, score-0.302]

99 Chaos in wavelength with a feedback tunable laser diode. [sent-308, score-0.573]

100 A CMOS a ı a implementation of FitzHugh-Nagumo neuron model. [sent-326, score-0.174]

similar papers computed by tfidf model

tfidf for this paper:

wordName wordTfidf (topN-words)

[('optical', 0.468), ('pulse', 0.25), ('optoelectronic', 0.248), ('driver', 0.216), ('wavelength', 0.215), ('optics', 0.196), ('laser', 0.18), ('electronic', 0.154), ('voltage', 0.151), ('detected', 0.119), ('birefringent', 0.113), ('interconnection', 0.113), ('lasers', 0.113), ('pulsed', 0.11), ('spiking', 0.109), ('modulation', 0.108), ('implementation', 0.105), ('feedback', 0.099), ('coupling', 0.094), ('dynamical', 0.093), ('oscillators', 0.09), ('pulsing', 0.09), ('vcsel', 0.09), ('electronics', 0.089), ('nonlinear', 0.084), ('recovery', 0.082), ('timing', 0.079), ('tunable', 0.079), ('crystal', 0.078), ('excitatory', 0.076), ('chaos', 0.072), ('cavity', 0.072), ('spike', 0.069), ('neuron', 0.069), ('integrators', 0.068), ('rad', 0.068), ('retardation', 0.068), ('semiconductor', 0.068), ('sinusoidal', 0.067), ('numeric', 0.063), ('bifurcation', 0.063), ('driving', 0.062), ('detector', 0.06), ('trajectory', 0.058), ('advance', 0.057), ('excitability', 0.054), ('input', 0.053), ('circuits', 0.052), ('diode', 0.05), ('material', 0.05), ('power', 0.05), ('spikes', 0.048), ('response', 0.048), ('goedgebuer', 0.045), ('incident', 0.045), ('nagumo', 0.045), ('nullcline', 0.045), ('polarizer', 0.045), ('psaltis', 0.045), ('resonant', 0.045), ('romariz', 0.045), ('vcsels', 0.045), ('excitation', 0.045), ('hop', 0.043), ('waveforms', 0.041), ('pulses', 0.041), ('periodic', 0.041), ('implementing', 0.04), ('variation', 0.039), ('simulated', 0.039), ('latest', 0.039), ('wagner', 0.039), ('parallelism', 0.039), ('signal', 0.039), ('implementations', 0.038), ('cycle', 0.038), ('experimental', 0.037), ('system', 0.037), ('arrangement', 0.036), ('fn', 0.036), ('neurons', 0.036), ('output', 0.036), ('implement', 0.035), ('nonlinearity', 0.034), ('nerve', 0.033), ('carrier', 0.033), ('injection', 0.033), ('polarization', 0.033), ('phase', 0.033), ('modi', 0.033), ('veri', 0.032), ('coupled', 0.032), ('transfer', 0.032), ('stability', 0.032), ('operation', 0.032), ('equilibrium', 0.032), ('completes', 0.031), ('thermal', 0.031), ('exhibits', 0.03), ('usa', 0.03)]

similar papers list:

simIndex simValue paperId paperTitle

same-paper 1 0.99999976 160 nips-2002-Optoelectronic Implementation of a FitzHugh-Nagumo Neural Model

Author: Alexandre R. Romariz, Kelvin Wagner

2 0.15271498 51 nips-2002-Classifying Patterns of Visual Motion - a Neuromorphic Approach

Author: Jakob Heinzle, Alan Stocker

Abstract: We report a system that classiﬁes and can learn to classify patterns of visual motion on-line. The complete system is described by the dynamics of its physical network architectures. The combination of the following properties makes the system novel: Firstly, the front-end of the system consists of an aVLSI optical ﬂow chip that collectively computes 2-D global visual motion in real-time [1]. Secondly, the complexity of the classiﬁcation task is signiﬁcantly reduced by mapping the continuous motion trajectories to sequences of ’motion events’. And thirdly, all the network structures are simple and with the exception of the optical ﬂow chip based on a Winner-Take-All (WTA) architecture. We demonstrate the application of the proposed generic system for a contactless man-machine interface that allows to write letters by visual motion. Regarding the low complexity of the system, its robustness and the already existing front-end, a complete aVLSI system-on-chip implementation is realistic, allowing various applications in mobile electronic devices.

3 0.12876265 76 nips-2002-Dynamical Constraints on Computing with Spike Timing in the Cortex

Author: Arunava Banerjee, Alexandre Pouget

Abstract: If the cortex uses spike timing to compute, the timing of the spikes must be robust to perturbations. Based on a recent framework that provides a simple criterion to determine whether a spike sequence produced by a generic network is sensitive to initial conditions, and numerical simulations of a variety of network architectures, we argue within the limits set by our model of the neuron, that it is unlikely that precise sequences of spike timings are used for computation under conditions typically found in the cortex.

4 0.11008598 50 nips-2002-Circuit Model of Short-Term Synaptic Dynamics

Author: Shih-Chii Liu, Malte Boegershausen, Pascal Suter

Abstract: We describe a model of short-term synaptic depression that is derived from a silicon circuit implementation. The dynamics of this circuit model are similar to the dynamics of some present theoretical models of shortterm depression except that the recovery dynamics of the variable describing the depression is nonlinear and it also depends on the presynaptic frequency. The equations describing the steady-state and transient responses of this synaptic model ﬁt the experimental results obtained from a fabricated silicon network consisting of leaky integrate-and-ﬁre neurons and different types of synapses. We also show experimental data demonstrating the possible computational roles of depression. One possible role of a depressing synapse is that the input can quickly bring the neuron up to threshold when the membrane potential is close to the resting potential.

5 0.10736832 171 nips-2002-Reconstructing Stimulus-Driven Neural Networks from Spike Times

Author: Duane Q. Nykamp

Abstract: We present a method to distinguish direct connections between two neurons from common input originating from other, unmeasured neurons. The distinction is computed from the spike times of the two neurons in response to a white noise stimulus. Although the method is based on a highly idealized linear-nonlinear approximation of neural response, we demonstrate via simulation that the approach can work with a more realistic, integrate-and-ﬁre neuron model. We propose that the approach exempliﬁed by this analysis may yield viable tools for reconstructing stimulus-driven neural networks from data gathered in neurophysiology experiments.

6 0.099106841 154 nips-2002-Neuromorphic Bisable VLSI Synapses with Spike-Timing-Dependent Plasticity

7 0.098590083 5 nips-2002-A Digital Antennal Lobe for Pattern Equalization: Analysis and Design

8 0.090444662 177 nips-2002-Retinal Processing Emulation in a Programmable 2-Layer Analog Array Processor CMOS Chip

9 0.082003899 43 nips-2002-Binary Coding in Auditory Cortex

10 0.080637917 11 nips-2002-A Model for Real-Time Computation in Generic Neural Microcircuits

11 0.079109333 129 nips-2002-Learning in Spiking Neural Assemblies

12 0.073759094 141 nips-2002-Maximally Informative Dimensions: Analyzing Neural Responses to Natural Signals

13 0.072806902 184 nips-2002-Spectro-Temporal Receptive Fields of Subthreshold Responses in Auditory Cortex

14 0.071134783 196 nips-2002-The RA Scanner: Prediction of Rheumatoid Joint Inflammation Based on Laser Imaging

15 0.07079839 103 nips-2002-How Linear are Auditory Cortical Responses?

16 0.06836421 128 nips-2002-Learning a Forward Model of a Reflex

17 0.066956364 186 nips-2002-Spike Timing-Dependent Plasticity in the Address Domain

18 0.066101022 180 nips-2002-Selectivity and Metaplasticity in a Unified Calcium-Dependent Model

19 0.06510891 148 nips-2002-Morton-Style Factorial Coding of Color in Primary Visual Cortex

20 0.065030649 44 nips-2002-Binary Tuning is Optimal for Neural Rate Coding with High Temporal Resolution

similar papers computed by lsi model

lsi for this paper:

topicId topicWeight

[(0, -0.159), (1, 0.184), (2, -0.004), (3, -0.046), (4, 0.044), (5, 0.102), (6, 0.041), (7, -0.005), (8, 0.051), (9, 0.036), (10, -0.014), (11, 0.086), (12, 0.003), (13, -0.011), (14, 0.005), (15, -0.017), (16, 0.012), (17, -0.036), (18, 0.033), (19, -0.043), (20, 0.088), (21, 0.033), (22, 0.041), (23, -0.025), (24, 0.07), (25, 0.062), (26, 0.012), (27, 0.038), (28, 0.039), (29, 0.059), (30, 0.047), (31, -0.049), (32, -0.078), (33, 0.038), (34, 0.0), (35, 0.067), (36, -0.046), (37, 0.141), (38, -0.048), (39, -0.087), (40, 0.067), (41, -0.137), (42, -0.049), (43, 0.126), (44, 0.016), (45, 0.011), (46, 0.0), (47, 0.033), (48, 0.03), (49, -0.027)]

similar papers list:

simIndex simValue paperId paperTitle

same-paper 1 0.96163523 160 nips-2002-Optoelectronic Implementation of a FitzHugh-Nagumo Neural Model

Author: Alexandre R. Romariz, Kelvin Wagner

2 0.73623431 5 nips-2002-A Digital Antennal Lobe for Pattern Equalization: Analysis and Design

Author: Alex Holub, Gilles Laurent, Pietro Perona

Abstract: Re-mapping patterns in order to equalize their distribution may greatly simplify both the structure and the training of classifiers. Here, the properties of one such map obtained by running a few steps of discrete-time dynamical system are explored. The system is called 'Digital Antennal Lobe' (DAL) because it is inspired by recent studies of the antennallobe, a structure in the olfactory system of the grasshopper. The pattern-spreading properties of the DAL as well as its average behavior as a function of its (few) design parameters are analyzed by extending previous results of Van Vreeswijk and Sompolinsky. Furthermore, a technique for adapting the parameters of the initial design in order to obtain opportune noise-rejection behavior is suggested. Our results are demonstrated with a number of simulations. 1

3 0.60588664 50 nips-2002-Circuit Model of Short-Term Synaptic Dynamics

Author: Shih-Chii Liu, Malte Boegershausen, Pascal Suter

4 0.60524493 11 nips-2002-A Model for Real-Time Computation in Generic Neural Microcircuits

Author: Wolfgang Maass, Thomas Natschläger, Henry Markram

Abstract: A key challenge for neural modeling is to explain how a continuous stream of multi-modal input from a rapidly changing environment can be processed by stereotypical recurrent circuits of integrate-and-ﬁre neurons in real-time. We propose a new computational model that is based on principles of high dimensional dynamical systems in combination with statistical learning theory. It can be implemented on generic evolved or found recurrent circuitry.

5 0.58971494 22 nips-2002-Adaptive Nonlinear System Identification with Echo State Networks

Author: Herbert Jaeger

Abstract: Echo state networks (ESN) are a novel approach to recurrent neural network training. An ESN consists of a large, fixed, recurrent

6 0.58344078 51 nips-2002-Classifying Patterns of Visual Motion - a Neuromorphic Approach

7 0.55997628 171 nips-2002-Reconstructing Stimulus-Driven Neural Networks from Spike Times

8 0.53842056 128 nips-2002-Learning a Forward Model of a Reflex

9 0.49962869 177 nips-2002-Retinal Processing Emulation in a Programmable 2-Layer Analog Array Processor CMOS Chip

10 0.49650159 154 nips-2002-Neuromorphic Bisable VLSI Synapses with Spike-Timing-Dependent Plasticity

11 0.4917573 76 nips-2002-Dynamical Constraints on Computing with Spike Timing in the Cortex

12 0.48432407 91 nips-2002-Field-Programmable Learning Arrays

13 0.46584418 71 nips-2002-Dopamine Induced Bistability Enhances Signal Processing in Spiny Neurons

14 0.46369502 23 nips-2002-Adaptive Quantization and Density Estimation in Silicon

15 0.45624396 44 nips-2002-Binary Tuning is Optimal for Neural Rate Coding with High Temporal Resolution

16 0.4335016 12 nips-2002-A Neural Edge-Detection Model for Enhanced Auditory Sensitivity in Modulated Noise

17 0.42179322 123 nips-2002-Learning Attractor Landscapes for Learning Motor Primitives

18 0.3970826 180 nips-2002-Selectivity and Metaplasticity in a Unified Calcium-Dependent Model

19 0.37782657 186 nips-2002-Spike Timing-Dependent Plasticity in the Address Domain

20 0.37624234 43 nips-2002-Binary Coding in Auditory Cortex

similar papers computed by lda model

lda for this paper:

topicId topicWeight

[(6, 0.011), (23, 0.031), (42, 0.04), (54, 0.079), (55, 0.043), (57, 0.014), (63, 0.357), (67, 0.04), (68, 0.07), (74, 0.067), (83, 0.02), (92, 0.023), (98, 0.107)]

similar papers list:

simIndex simValue paperId paperTitle

same-paper 1 0.81943256 160 nips-2002-Optoelectronic Implementation of a FitzHugh-Nagumo Neural Model

Author: Alexandre R. Romariz, Kelvin Wagner

2 0.58342505 188 nips-2002-Stability-Based Model Selection

Author: Tilman Lange, Mikio L. Braun, Volker Roth, Joachim M. Buhmann

Abstract: Model selection is linked to model assessment, which is the problem of comparing different models, or model parameters, for a speciﬁc learning task. For supervised learning, the standard practical technique is crossvalidation, which is not applicable for semi-supervised and unsupervised settings. In this paper, a new model assessment scheme is introduced which is based on a notion of stability. The stability measure yields an upper bound to cross-validation in the supervised case, but extends to semi-supervised and unsupervised problems. In the experimental part, the performance of the stability measure is studied for model order selection in comparison to standard techniques in this area.

3 0.4301737 11 nips-2002-A Model for Real-Time Computation in Generic Neural Microcircuits

Author: Wolfgang Maass, Thomas Natschläger, Henry Markram

4 0.42939085 5 nips-2002-A Digital Antennal Lobe for Pattern Equalization: Analysis and Design

Author: Alex Holub, Gilles Laurent, Pietro Perona

5 0.42502987 62 nips-2002-Coulomb Classifiers: Generalizing Support Vector Machines via an Analogy to Electrostatic Systems

Author: Sepp Hochreiter, Michael C. Mozer, Klaus Obermayer

Abstract: We introduce a family of classiﬁers based on a physical analogy to an electrostatic system of charged conductors. The family, called Coulomb classiﬁers, includes the two best-known support-vector machines (SVMs), the ν–SVM and the C–SVM. In the electrostatics analogy, a training example corresponds to a charged conductor at a given location in space, the classiﬁcation function corresponds to the electrostatic potential function, and the training objective function corresponds to the Coulomb energy. The electrostatic framework provides not only a novel interpretation of existing algorithms and their interrelationships, but it suggests a variety of new methods for SVMs including kernels that bridge the gap between polynomial and radial-basis functions, objective functions that do not require positive-deﬁnite kernels, regularization techniques that allow for the construction of an optimal classiﬁer in Minkowski space. Based on the framework, we propose novel SVMs and perform simulation studies to show that they are comparable or superior to standard SVMs. The experiments include classiﬁcation tasks on data which are represented in terms of their pairwise proximities, where a Coulomb Classiﬁer outperformed standard SVMs. 1

6 0.42499912 76 nips-2002-Dynamical Constraints on Computing with Spike Timing in the Cortex

7 0.41939184 148 nips-2002-Morton-Style Factorial Coding of Color in Primary Visual Cortex

8 0.41885567 141 nips-2002-Maximally Informative Dimensions: Analyzing Neural Responses to Natural Signals

9 0.41881964 44 nips-2002-Binary Tuning is Optimal for Neural Rate Coding with High Temporal Resolution

10 0.41748086 50 nips-2002-Circuit Model of Short-Term Synaptic Dynamics

11 0.41747665 43 nips-2002-Binary Coding in Auditory Cortex

12 0.41691864 73 nips-2002-Dynamic Bayesian Networks with Deterministic Latent Tables

13 0.41681075 28 nips-2002-An Information Theoretic Approach to the Functional Classification of Neurons

14 0.41675329 81 nips-2002-Expected and Unexpected Uncertainty: ACh and NE in the Neocortex

15 0.41535378 102 nips-2002-Hidden Markov Model of Cortical Synaptic Plasticity: Derivation of the Learning Rule

16 0.41530216 199 nips-2002-Timing and Partial Observability in the Dopamine System

17 0.41335356 10 nips-2002-A Model for Learning Variance Components of Natural Images

18 0.41130722 123 nips-2002-Learning Attractor Landscapes for Learning Motor Primitives

19 0.41063166 184 nips-2002-Spectro-Temporal Receptive Fields of Subthreshold Responses in Auditory Cortex

20 0.40833816 51 nips-2002-Classifying Patterns of Visual Motion - a Neuromorphic Approach