nips nips2011 nips2011-250 knowledge-graph by maker-knowledge-mining

250 nips-2011-Shallow vs. Deep Sum-Product Networks

Source: pdf

Author: Olivier Delalleau, Yoshua Bengio

Abstract: We investigate the representational power of sum-product networks (computation networks analogous to neural networks, but whose individual units compute either products or weighted sums), through a theoretical analysis that compares deep (multiple hidden layers) vs. shallow (one hidden layer) architectures. We prove there exist families of functions that can be represented much more efﬁciently with a deep network than with a shallow one, i.e. with substantially fewer hidden units. Such results were not available until now, and contribute to motivate recent research involving learning of deep sum-product networks, and more generally motivate research in Deep Learning. 1 Introduction and prior work Many learning algorithms are based on searching a family of functions so as to identify one member of said family which minimizes a training criterion. The choice of this family of functions and how members of that family are parameterized can be a crucial one. Although there is no universally optimal choice of parameterization or family of functions (or “architecture”), as demonstrated by the no-free-lunch results [37], it may be the case that some architectures are appropriate (or inappropriate) for a large class of learning tasks and data distributions, such as those related to Artiﬁcial Intelligence (AI) tasks [4]. Different families of functions have different characteristics that can be appropriate or not depending on the learning task of interest. One of the characteristics that has spurred much interest and research in recent years is depth of the architecture. In the case of a multi-layer neural network, depth corresponds to the number of (hidden and output) layers. A ﬁxedkernel Support Vector Machine is considered to have depth 2 [4] and boosted decision trees to have depth 3 [7]. Here we use the word circuit or network to talk about a directed acyclic graph, where each node is associated with some output value which can be computed based on the values associated with its predecessor nodes. The arguments of the learned function are set at the input nodes of the circuit (which have no predecessor) and the outputs of the function are read off the output nodes of the circuit. Different families of functions correspond to different circuits and allowed choices of computations in each node. Learning can be performed by changing the computation associated with a node, or rewiring the circuit (possibly changing the number of nodes). The depth of the circuit is the length of the longest path in the graph from an input node to an output node. Deep Learning algorithms [3] are tailored to learning circuits with variable depth, typically greater than depth 2. They are based on the idea of multiple levels of representation, with the intuition that the raw input can be represented at different levels of abstraction, with more abstract features of the input or more abstract explanatory factors represented by deeper circuits. These algorithms are often based on unsupervised learning, opening the door to semi-supervised learning and efﬁcient 1 use of large quantities of unlabeled data [3]. Analogies with the structure of the cerebral cortex (in particular the visual cortex) [31] and similarities between features learned with some Deep Learning algorithms and those hypothesized in the visual cortex [17] further motivate investigations into deep architectures. It has been suggested that deep architectures are more powerful in the sense of being able to more efﬁciently represent highly-varying functions [4, 3]. In this paper, we measure “efﬁciency” in terms of the number of computational units in the network. An efﬁcient representation is important mainly because: (i) it uses less memory and is faster to compute, and (ii) given a ﬁxed amount of training samples and computational power, better generalization is expected. The ﬁrst successful algorithms for training deep architectures appeared in 2006, with efﬁcient training procedures for Deep Belief Networks [14] and deep auto-encoders [13, 27, 6], both exploiting the general idea of greedy layer-wise pre-training [6]. Since then, these ideas have been investigated further and applied in many settings, demonstrating state-of-the-art learning performance in object recognition [16, 28, 18, 15] and segmentation [20], audio classiﬁcation [19, 10], natural language processing [9, 36, 21, 32], collaborative ﬁltering [30], modeling textures [24], modeling motion [34, 33], information retrieval [29, 26], and semi-supervised learning [36, 22]. Poon and Domingos [25] introduced deep sum-product networks as a method to compute partition functions of tractable graphical models. These networks are analogous to traditional artiﬁcial neural networks but with nodes that compute either products or weighted sums of their inputs. Analogously to neural networks, we deﬁne “hidden” nodes as those nodes that are neither input nodes nor output nodes. If the nodes are organized in layers, we deﬁne the “hidden” layers to be those that are neither the input layer nor the output layer. Poon and Domingos [25] report experiments with networks much deeper (30+ hidden layers) than those typically used until now, e.g. in Deep Belief Networks [14, 3], where the number of hidden layers is usually on the order of three to ﬁve. Whether such deep architectures have theoretical advantages compared to so-called “shallow” architectures (i.e. those with a single hidden layer) remains an open question. After all, in the case of a sum-product network, the output value can always be written as a sum of products of input variables (possibly raised to some power by allowing multiple connections from the same input), and consequently it is easily rewritten as a shallow network with a sum output unit and product hidden units. The argument supported by our theoretical analysis is that a deep architecture is able to compute some functions much more efﬁciently than a shallow one. Until recently, very few theoretical results supported the idea that deep architectures could present an advantage in terms of representing some functions more efﬁciently. Most related results originate from the analysis of boolean circuits (see e.g. [2] for a review). Well-known results include the proof that solving the n-bit parity task with a depth-2 circuit requires an exponential number of gates [1, 38], and more generally that there exist functions computable with a polynomial-size depthk circuit that would require exponential size when restricted to depth k − 1 [11]. Another recent result on boolean circuits by Braverman [8] offers proof of a longstanding conjecture, showing that bounded-depth boolean circuits are unable to distinguish some (non-uniform) input distributions from the uniform distribution (i.e. they are “fooled” by such input distributions). In particular, Braverman’s result suggests that shallow circuits can in general be fooled more easily than deep ones, i.e., that they would have more difﬁculty efﬁciently representing high-order dependencies (those involving many input variables). It is not obvious that circuit complexity results (that typically consider only boolean or at least discrete nodes) are directly applicable in the context of typical machine learning algorithms such as neural networks (that compute continuous representations of their input). Orponen [23] surveys theoretical results in computational complexity that are relevant to learning algorithms. For instance, H˚ stad and Goldmann [12] extended some results to the case of networks of linear threshold units a with positivity constraints on the weights. Bengio et al. [5, 7] investigate, respectively, complexity issues in networks of Gaussian radial basis functions and decision trees, showing intrinsic limitations of these architectures e.g. on tasks similar to the parity problem. Utgoff and Stracuzzi [35] informally discuss the advantages of depth in boolean circuit in the context of learning architectures. Bengio [3] suggests that some polynomials could be represented more efﬁciently by deep sumproduct networks, but without providing any formal statement or proofs. This work partly addresses this void by demonstrating families of circuits for which a deep architecture can be exponentially more efﬁcient than a shallow one in the context of real-valued polynomials. Note that we do not address in this paper the problem of learning these parameters: even if an efﬁcient deep representation exists for the function we seek to approximate, in general there is no 2 guarantee for standard optimization algorithms to easily converge to this representation. This paper focuses on the representational power of deep sum-product circuits compared to shallow ones, and studies it by considering particular families of target functions (to be represented by the learner). We ﬁrst formally deﬁne sum-product networks. We consider two families of functions represented by deep sum-product networks (families F and G). For each family, we establish a lower bound on the minimal number of hidden units a depth-2 sum-product network would require to represent a function of this family, showing it is much less efﬁcient than the deep representation. 2 Sum-product networks Deﬁnition 1. A sum-product network is a network composed of units that either compute the product of their inputs or a weighted sum of their inputs (where weights are strictly positive). Here, we restrict our deﬁnition of the generic term “sum-product network” to networks whose summation units have positive incoming weights1 , while others are called “negative-weight” networks. Deﬁnition 2. A “negative-weight“ sum-product network may contain summation units whose weights are non-positive (i.e. less than or equal to zero). Finally, we formally deﬁne what we mean by deep vs. shallow networks in the rest of the paper. Deﬁnition 3. A “shallow“ sum-product network contains a single hidden layer (i.e. a total of three layers when counting the input and output layers, and a depth equal to two). Deﬁnition 4. A “deep“ sum-product network contains more than one hidden layer (i.e. a total of at least four layers, and a depth at least three). The family F 3 3.1 Deﬁnition The ﬁrst family of functions we study, denoted by F, is made of functions built from deep sumproduct networks that alternate layers of product and sum units with two inputs each (details are provided below). The basic idea we use here is that composing layers (i.e. using a deep architecture) is equivalent to using a factorized representation of the polynomial function computed by the network. Such a factorized representation can be exponentially more compact than its expansion as a sum of products (which can be associated to a shallow network with product units in its hidden layer and a sum unit as output). This is what we formally show in what follows. + ℓ2 = λ11ℓ1 + µ11ℓ1 = x1x2 + x3x4 = f (x1, x2, x3, x4) 2 1 1 λ11 = 1 µ11 = 1 × ℓ1 = x1x2 1 x1 x2 × ℓ1 = x3x4 2 x3 x4 Figure 1: Sum-product network computing the function f ∈ F such that i = λ11 = µ11 = 1. Let n = 4i , with i a positive integer value. Denote by ℓ0 the input layer containing scalar variables {x1 , . . . , xn }, such that ℓ0 = xj for 1 ≤ j ≤ n. Now deﬁne f ∈ F as any function computed by a j sum-product network (deep for i ≥ 2) composed of alternating product and sum layers: • ℓ2k+1 = ℓ2k · ℓ2k for 0 ≤ k ≤ i − 1 and 1 ≤ j ≤ 22(i−k)−1 2j−1 2j j • ℓ2k = λjk ℓ2k−1 + µjk ℓ2k−1 for 1 ≤ k ≤ i and 1 ≤ j ≤ 22(i−k) j 2j 2j−1 where the weights λjk and µjk of the summation units are strictly positive. The output of the network is given by f (x1 , . . . , xn ) = ℓ2i ∈ R, the unique unit in the last layer. 1 The corresponding (shallow) network for i = 1 and additive weights set to one is shown in Figure 1 1 This condition is required by some of the proofs presented here. 3 (this architecture is also the basic building block of bigger networks for i > 1). Note that both the input size n = 4i and the network’s depth 2i increase with parameter i. 3.2 Theoretical results The main result of this section is presented below in Corollary 1, providing a lower bound on the minimum number of hidden units required by a shallow sum-product network to represent a function f ∈ F. The high-level proof sketch consists in the following steps: (1) Count the number of unique products found in the polynomial representation of f (Lemma 1 and Proposition 1). (2) Show that the only possible architecture for a shallow sum-product network to compute f is to have a hidden layer made of product units, with a sum unit as output (Lemmas 2 to 5). (3) Conclude that the number of hidden units must be at least the number of unique products computed in step 3.2 (Lemma 6 and Corollary 1). Lemma 1. Any element ℓk can be written as a (positively) weighted sum of products of input varij ables, such that each input variable xt is used in exactly one unit of ℓk . Moreover, the number mk of products found in the sum computed by ℓk does not depend on j and obeys the following recurrence j rule for k ≥ 0: if k + 1 is odd, then mk+1 = m2 , otherwise mk+1 = 2mk . k Proof. We prove the lemma by induction on k. It is obviously true for k = 0 since ℓ0 = xj . j Assuming this is true for some k ≥ 0, we consider two cases: k+1 k • If k + 1 is odd, then ℓj = ℓk 2j−1 · ℓ2j . By the inductive hypothesis, it is the product of two (positively) weighted sums of products of input variables, and no input variable can k appear in both ℓk 2j−1 and ℓ2j , so the result is also a (positively) weighted sum of products k of input variables. Additionally, if the number of products in ℓk 2j−1 and ℓ2j is mk , then 2 mk+1 = mk , since all products involved in the multiplication of the two units are different (since they use disjoint subsets of input variables), and the sums have positive weights. Finally, by the induction assumption, an input variable appears in exactly one unit of ℓk . This unit is an input to a single unit of ℓk+1 , that will thus be the only unit of ℓk+1 where this input variable appears. k • If k + 1 is even, then ℓk+1 = λjk ℓk 2j−1 + µjk ℓ2j . Again, from the induction assumption, it j must be a (positively) weighted sum of products of input variables, but with mk+1 = 2mk such products. As in the previous case, an input variable will appear in the single unit of ℓk+1 that has as input the single unit of ℓk in which this variable must appear. 2i Proposition 1. The number of products in the sum computed in the output unit l1 of a network √ n−1 . computing a function in F is m2i = 2 Proof. We ﬁrst prove by induction on k ≥ 1 that for odd k, mk = 22 k 22 1+1 2 2 k+1 2 −2 , and for even k, . This is obviously true for k = 1 since 2 = 2 = 1, and all units in ℓ1 are mk = 2 single products of the form xr xs . Assuming this is true for some k ≥ 1, then: −1 0 −2 • if k + 1 is odd, then from Lemma 1 and the induction assumption, we have: mk+1 = m2 = k 2 k 22 2 −1 k +1 = 22 2 • if k + 1 is even, then instead we have: mk+1 = 2mk = 2 · 22 k+1 2 −2 −2 = 22 = 22 (k+1)+1 2 (k+1) 2 −2 −1 which shows the desired result for k + 1, and thus concludes the induction proof. Applying this result with k = 2i (which is even) yields 2i m2i = 22 2 −1 √ =2 4 22i −1 √ =2 n−1 . 2i Lemma 2. The products computed in the output unit l1 can be split in two groups, one with products containing only variables x1 , . . . , x n and one containing only variables x n +1 , . . . , xn . 2 2 Proof. This is obvious since the last unit is a “sum“ unit that adds two terms whose inputs are these two groups of variables (see e.g. Fig. 1). 2i Lemma 3. The products computed in the output unit l1 involve more than one input variable. k Proof. It is straightforward to show by induction on k ≥ 1 that the products computed by lj all involve more than one input variable, thus it is true in particular for the output layer (k = 2i). Lemma 4. Any shallow sum-product network computing f ∈ F must have a “sum” unit as output. Proof. By contradiction, suppose the output unit of such a shallow sum-product network is multiplicative. This unit must have more than one input, because in the case that it has only one input, the output would be either a (weighted) sum of input variables (which would violate Lemma 3), or a single product of input variables (which would violate Proposition 1), depending on the type (sum or product) of the single input hidden unit. Thus the last unit must compute a product of two or more hidden units. It can be re-written as a product of two factors, where each factor corresponds to either one hidden unit, or a product of multiple hidden units (it does not matter here which speciﬁc factorization is chosen among all possible ones). Regardless of the type (sum or product) of the hidden units involved, those two factors can thus be written as weighted sums of products of variables xt (with positive weights, and input variables potentially raised to powers above one). From Lemma 1, both x1 and xn must be present in the ﬁnal output, and thus they must appear in at least one of these two factors. Without loss of generality, assume x1 appears in the ﬁrst factor. Variables x n +1 , . . . , xn then cannot be present in the second factor, since otherwise one product in the output 2 would contain both x1 and one of these variables (this product cannot cancel out since weights must be positive), violating Lemma 2. But with a similar reasoning, since as a result xn must appear in the ﬁrst factor, variables x1 , . . . , x n cannot be present in the second factor either. Consequently, no 2 input variable can be present in the second factor, leading to the desired contradiction. Lemma 5. Any shallow sum-product network computing f ∈ F must have only multiplicative units in its hidden layer. Proof. By contradiction, suppose there exists a “sum“ unit in the hidden layer, written s = t∈S αt xt with S the set of input indices appearing in this sum, and αt > 0 for all t ∈ S. Since according to Lemma 4 the output unit must also be a sum (and have positive weights according to Deﬁnition 1), then the ﬁnal output will also contain terms of the form βt xt for t ∈ S, with βt > 0. This violates Lemma 3, establishing the contradiction. Lemma 6. Any shallow negative-weight sum-product network (see Deﬁnition 2) computing f ∈ F √ must have at least 2 n−1 hidden units, if its output unit is a sum and its hidden units are products. Proof. Such a network computes a weighted sum of its hidden units, where each hidden unit is a γ product of input variables, i.e. its output can be written as Σj wj Πt xt jt with wj ∈ R and γjt ∈ {0, 1}. In order to compute a function in F, this shallow network thus needs a number of hidden units at least equal to the number of unique products in that function. From Proposition 1, this √ number is equal to 2 n−1 . √ Corollary 1. Any shallow sum-product network computing f ∈ F must have at least 2 units. n−1 hidden Proof. This is a direct corollary of Lemmas 4 (showing the output unit is a sum), 5 (showing that hidden units are products), and 6 (showing the desired result for any shallow network with this speciﬁc structure – regardless of the sign of weights). 5 3.3 Discussion Corollary 1 above shows that in order to compute some function in F with n inputs, the number of √ √ units in a shallow network has to be at least 2 n−1 , (i.e. grows exponentially in n). On another hand, the total number of units in the deep (for i > 1) network computing the same function, as described in Section 3.1, is equal to 1 + 2 + 4 + 8 + . . . + 22i−1 (since all units are binary), which is √ also equal to 22i − 1 = n − 1 (i.e. grows only quadratically in n). It shows that some deep sumproduct network with n inputs and depth O(log n) can represent with O(n) units what would √ require O(2 n ) units for a depth-2 network. Lemma 6 also shows a similar result regardless of the sign of the weights in the summation units of the depth-2 network, but assumes a speciﬁc architecture for this network (products in the hidden layer with a sum as output). 4 The family G In this section we present similar results with a different family of functions, denoted by G. Compared to F, one important difference of deep sum-product networks built to deﬁne functions in G is that they can vary their input size independently of their depth. Their analysis thus provides additional insight when comparing the representational efﬁciency of deep vs. shallow sum-product networks in the case of a ﬁxed dataset. 4.1 Deﬁnition Networks in family G also alternate sum and product layers, but their units have as inputs all units from the previous layer except one. More formally, deﬁne the family G = ∪n≥2,i≥0 Gin of functions represented by sum-product networks, where the sub-family Gin is made of all sum-product networks with n input variables and 2i + 2 layers (including the input layer ℓ0 ), such that: 1. ℓ1 contains summation units; further layers alternate multiplicative and summation units. 2. Summation units have positive weights. 3. All layers are of size n, except the last layer ℓ2i+1 that contains a single sum unit that sums all units in the previous layer ℓ2i . k−1 4. In each layer ℓk for 1 ≤ k ≤ 2i, each unit ℓk takes as inputs {ℓm |m = j}. j An example of a network belonging to G1,3 (i.e. with three layers and three input variables) is shown in Figure 2. ℓ3 = x2 + x2 + x2 + 3(x1x2 + x1x3 + x2x3) = g(x1, x2, x3) 3 2 1 1 + ℓ2 = x2 + x1x2 × 1 1 +x1x3 + x2x3 ℓ1 = x2 + x3 1 × ℓ2 = . . . 2 × ℓ2 = x2 + x1x2 3 3 +x1x3 + x2x3 + + ℓ1 = x1 + x3 2 + ℓ1 = x1 + x2 3 x1 x2 x3 Figure 2: Sum-product network computing a function of G1,3 (summation units’ weights are all 1’s). 4.2 Theoretical results The main result is stated in Proposition 3 below, establishing a lower bound on the number of hidden units of a shallow sum-product network computing g ∈ G. The proof sketch is as follows: 1. We show that the polynomial expansion of g must contain a large set of products (Proposition 2 and Corollary 2). 2. We use both the number of products in that set as well as their degree to establish the desired lower bound (Proposition 3). 6 We will also need the following lemma, which states that when n − 1 items each belong to n − 1 sets among a total of n sets, then we can associate to each item one of the sets it belongs to without using the same set for different items. Lemma 7. Let S1 , . . . , Sn be n sets (n ≥ 2) containing elements of {P1 , . . . , Pn−1 }, such that for any q, r, |{r|Pq ∈ Sr }| ≥ n − 1 (i.e. each element Pq belongs to at least n − 1 sets). Then there exist r1 , . . . , rn−1 different indices such that Pq ∈ Srq for 1 ≤ q ≤ n − 1. Proof. Omitted due to lack of space (very easy to prove by construction). Proposition 2. For any 0 ≤ j ≤ i, and any product of variables P = Πn xαt such that αt ∈ N and t=1 t j 2j whose computed value, when expanded as a weighted t αt = (n − 1) , there exists a unit in ℓ sum of products, contains P among these products. Proof. We prove this proposition by induction on j. First, for j = 0, this is obvious since any P of this form must be made of a single input variable xt , that appears in ℓ0 = xt . t Suppose now the proposition is true for some j < i. Consider a product P = Πn xαt such that t=1 t αt ∈ N and t αt = (n − 1)j+1 . P can be factored in n − 1 sub-products of degree (n − 1)j , β i.e. written P = P1 . . . Pn−1 with Pq = Πn xt qt , βqt ∈ N and t βqt = (n − 1)j for all q. By t=1 the induction hypothesis, each Pq can be found in at least one unit ℓ2j . As a result, by property 4 kq (in the deﬁnition of family G), each Pq will also appear in the additive layer ℓ2j+1 , in at least n − 1 different units (the only sum unit that may not contain Pq is the one that does not have ℓ2j as input). kq By Lemma 7, we can thus ﬁnd a set of units ℓ2j+1 such that for any 1 ≤ q ≤ n − 1, the product rq Pq appears in ℓ2j+1 , with indices rq being different from each other. Let 1 ≤ s ≤ n be such that rq 2(j+1) s = rq for all q. Then, from property 4 of family G, the multiplicative unit ℓs computes the n−1 2j+1 product Πq=1 ℓrq , and as a result, when expanded as a sum of products, it contains in particular P1 . . . Pn−1 = P . The proposition is thus true for j + 1, and by induction, is true for all j ≤ i. Corollary 2. The output gin of a sum-product network in Gin , when expanded as a sum of products, contains all products of variables of the form Πn xαt such that αt ∈ N and t αt = (n − 1)i . t=1 t Proof. Applying Proposition 2 with j = i, we obtain that all products of this form can be found in the multiplicative units of ℓ2i . Since the output unit ℓ2i+1 computes a sum of these multiplicative 1 units (weighted with positive weights), those products are also present in the output. Proposition 3. A shallow negative-weight sum-product network computing gin ∈ Gin must have at least (n − 1)i hidden units. Proof. First suppose the output unit of the shallow network is a sum. Then it may be able to compute gin , assuming we allow multiplicative units in the hidden layer in the hidden layer to use powers of their inputs in the product they compute (which we allow here for the proof to be more generic). However, it will require at least as many of these units as the number of unique products that can be found in the expansion of gin . In particular, from Corollary 2, it will require at least the number n of unique tuples of the form (α1 , . . . , αn ) such that αt ∈ N and t=1 αt = (n − 1)i . Denoting ni dni = (n − 1)i , this number is known to be equal to n+dni −1 , and it is easy to verify it is higher d than (or equal to) dni for any n ≥ 2 and i ≥ 0. Now suppose the output unit is multiplicative. Then there can be no multiplicative hidden unit, otherwise it would mean one could factor some input variable xt in the computed function output: this is not possible since by Corollary 2, for any variable xt there exist products in the output function that do not involve xt . So all hidden units must be additive, and since the computed function contains products of degree dni , there must be at least dni such hidden units. 7 4.3 Discussion Proposition 3 shows that in order to compute the same function as gin ∈ Gin , the number of units in the shallow network has to grow exponentially in i, i.e. in the network’s depth (while the deep network’s size grows linearly in i). The shallow network also needs to grow polynomially in the number of input variables n (with a degree equal to i), while the deep network grows only linearly in n. It means that some deep sum-product network with n inputs and depth O(i) can represent with O(ni) units what would require O((n − 1)i ) units for a depth-2 network. Note that in the similar results found for family F, the depth-2 network computing the same function as a function in F had to be constrained to either have a speciﬁc combination of sum and hidden units (in Lemma 6) or to have non-negative weights (in Corollary 1). On the contrary, the result presented here for family G holds without requiring any of these assumptions. 5 Conclusion We compared a deep sum-product network and a shallow sum-product network representing the same function, taken from two families of functions F and G. For both families, we have shown that the number of units in the shallow network has to grow exponentially, compared to a linear growth in the deep network, so as to represent the same functions. The deep version thus offers a much more compact representation of the same functions. This work focuses on two speciﬁc families of functions: ﬁnding more general parameterization of functions leading to similar results would be an interesting topic for future research. Another open question is whether it is possible to represent such functions only approximately (e.g. up to an error bound ǫ) with a much smaller shallow network. Results by Braverman [8] on boolean circuits suggest that similar results as those presented in this paper may still hold, but this topic has yet to be formally investigated in the context of sum-product networks. A related problem is also to look into functions deﬁned only on discrete input variables: our proofs do not trivially extend to this situation because we cannot assume anymore that two polynomials yielding the same output values must have the same expansion coefﬁcients (since the number of input combinations becomes ﬁnite). Acknowledgments The authors would like to thank Razvan Pascanu and David Warde-Farley for their help in improving this manuscript, as well as the anonymous reviewers for their careful reviews. This work was partially funded by NSERC, CIFAR, and the Canada Research Chairs. References [1] Ajtai, M. (1983). P1 1 -formulae on ﬁnite structures. Annals of Pure and Applied Logic, 24(1), 1–48. [2] Allender, E. (1996). Circuit complexity before the dawn of the new millennium. In 16th Annual Conference on Foundations of Software Technology and Theoretical Computer Science, pages 1–18. Lecture Notes in Computer Science 1180, Springer Verlag. [3] Bengio, Y. (2009). Learning deep architectures for AI. Foundations and Trends in Machine Learning, 2(1), 1–127. Also published as a book. Now Publishers, 2009. [4] Bengio, Y. and LeCun, Y. (2007). Scaling learning algorithms towards AI. In L. Bottou, O. Chapelle, D. DeCoste, and J. Weston, editors, Large Scale Kernel Machines. MIT Press. [5] Bengio, Y., Delalleau, O., and Le Roux, N. (2006). The curse of highly variable functions for local kernel machines. In NIPS’05, pages 107–114. MIT Press, Cambridge, MA. [6] Bengio, Y., Lamblin, P., Popovici, D., and Larochelle, H. (2007). Greedy layer-wise training of deep networks. In NIPS 19, pages 153–160. MIT Press. [7] Bengio, Y., Delalleau, O., and Simard, C. (2010). Decision trees do not generalize to new variations. Computational Intelligence, 26(4), 449–467. [8] Braverman, M. (2011). Poly-logarithmic independence fools bounded-depth boolean circuits. Communications of the ACM, 54(4), 108–115. [9] Collobert, R. and Weston, J. (2008). A uniﬁed architecture for natural language processing: Deep neural networks with multitask learning. In ICML 2008, pages 160–167. [10] Dahl, G. E., Ranzato, M., Mohamed, A., and Hinton, G. E. (2010). Phone recognition with the meancovariance restricted boltzmann machine. In Advances in Neural Information Processing Systems (NIPS). 8 [11] H˚ stad, J. (1986). Almost optimal lower bounds for small depth circuits. In Proceedings of the 18th a annual ACM Symposium on Theory of Computing, pages 6–20, Berkeley, California. ACM Press. [12] H˚ stad, J. and Goldmann, M. (1991). On the power of small-depth threshold circuits. Computational a Complexity, 1, 113–129. [13] Hinton, G. E. and Salakhutdinov, R. (2006). Reducing the dimensionality of data with neural networks. Science, 313(5786), 504–507. [14] Hinton, G. E., Osindero, S., and Teh, Y. (2006). A fast learning algorithm for deep belief nets. Neural Computation, 18, 1527–1554. [15] Kavukcuoglu, K., Sermanet, P., Boureau, Y.-L., Gregor, K., Mathieu, M., and LeCun, Y. (2010). Learning convolutional feature hierarchies for visual recognition. In NIPS’10. [16] Larochelle, H., Erhan, D., Courville, A., Bergstra, J., and Bengio, Y. (2007). An empirical evaluation of deep architectures on problems with many factors of variation. In ICML’07, pages 473–480. ACM. [17] Lee, H., Ekanadham, C., and Ng, A. (2008). Sparse deep belief net model for visual area V2. In NIPS’07, pages 873–880. MIT Press, Cambridge, MA. [18] Lee, H., Grosse, R., Ranganath, R., and Ng, A. Y. (2009a). Convolutional deep belief networks for scalable unsupervised learning of hierarchical representations. In ICML 2009. Montreal (Qc), Canada. [19] Lee, H., Pham, P., Largman, Y., and Ng, A. (2009b). Unsupervised feature learning for audio classiﬁcation using convolutional deep belief networks. In NIPS’09, pages 1096–1104. [20] Levner, I. (2008). Data Driven Object Segmentation. Ph.D. thesis, Department of Computer Science, University of Alberta. [21] Mnih, A. and Hinton, G. E. (2009). A scalable hierarchical distributed language model. In NIPS’08, pages 1081–1088. [22] Mobahi, H., Collobert, R., and Weston, J. (2009). Deep learning from temporal coherence in video. In ICML’2009, pages 737–744. [23] Orponen, P. (1994). Computational complexity of neural networks: a survey. Nordic Journal of Computing, 1(1), 94–110. [24] Osindero, S. and Hinton, G. E. (2008). Modeling image patches with a directed hierarchy of markov random ﬁeld. In NIPS’07, pages 1121–1128, Cambridge, MA. MIT Press. [25] Poon, H. and Domingos, P. (2011). Sum-product networks: A new deep architecture. In UAI’2011, Barcelona, Spain. [26] Ranzato, M. and Szummer, M. (2008). Semi-supervised learning of compact document representations with deep networks. In ICML. [27] Ranzato, M., Poultney, C., Chopra, S., and LeCun, Y. (2007). Efﬁcient learning of sparse representations with an energy-based model. In NIPS’06, pages 1137–1144. MIT Press. [28] Ranzato, M., Boureau, Y.-L., and LeCun, Y. (2008). Sparse feature learning for deep belief networks. In NIPS’07, pages 1185–1192, Cambridge, MA. MIT Press. [29] Salakhutdinov, R. and Hinton, G. E. (2007). Semantic hashing. In Proceedings of the 2007 Workshop on Information Retrieval and applications of Graphical Models (SIGIR 2007), Amsterdam. Elsevier. [30] Salakhutdinov, R., Mnih, A., and Hinton, G. E. (2007). Restricted Boltzmann machines for collaborative ﬁltering. In ICML 2007, pages 791–798, New York, NY, USA. [31] Serre, T., Kreiman, G., Kouh, M., Cadieu, C., Knoblich, U., and Poggio, T. (2007). A quantitative theory of immediate visual recognition. Progress in Brain Research, Computational Neuroscience: Theoretical Insights into Brain Function, 165, 33–56. [32] Socher, R., Lin, C., Ng, A. Y., and Manning, C. (2011). Learning continuous phrase representations and syntactic parsing with recursive neural networks. In ICML’2011. [33] Taylor, G. and Hinton, G. (2009). Factored conditional restricted Boltzmann machines for modeling motion style. In ICML 2009, pages 1025–1032. [34] Taylor, G., Hinton, G. E., and Roweis, S. (2007). Modeling human motion using binary latent variables. In NIPS’06, pages 1345–1352. MIT Press, Cambridge, MA. [35] Utgoff, P. E. and Stracuzzi, D. J. (2002). Many-layered learning. Neural Computation, 14, 2497–2539. [36] Weston, J., Ratle, F., and Collobert, R. (2008). Deep learning via semi-supervised embedding. In ICML 2008, pages 1168–1175, New York, NY, USA. [37] Wolpert, D. H. (1996). The lack of a priori distinction between learning algorithms. Neural Computation, 8(7), 1341–1390. [38] Yao, A. (1985). Separating the polynomial-time hierarchy by oracles. In Proceedings of the 26th Annual IEEE Symposium on Foundations of Computer Science, pages 1–10. 9

Reference: text

Summary: the most important sentenses genereted by tfidf model

sentIndex sentText sentNum sentScore

1 ca Abstract We investigate the representational power of sum-product networks (computation networks analogous to neural networks, but whose individual units compute either products or weighted sums), through a theoretical analysis that compares deep (multiple hidden layers) vs. [sent-6, score-1.467]

2 We prove there exist families of functions that can be represented much more efﬁciently with a deep network than with a shallow one, i. [sent-8, score-1.148]

3 Such results were not available until now, and contribute to motivate recent research involving learning of deep sum-product networks, and more generally motivate research in Deep Learning. [sent-11, score-0.454]

4 Here we use the word circuit or network to talk about a directed acyclic graph, where each node is associated with some output value which can be computed based on the values associated with its predecessor nodes. [sent-19, score-0.423]

5 The depth of the circuit is the length of the longest path in the graph from an input node to an output node. [sent-23, score-0.401]

6 Analogies with the structure of the cerebral cortex (in particular the visual cortex) [31] and similarities between features learned with some Deep Learning algorithms and those hypothesized in the visual cortex [17] further motivate investigations into deep architectures. [sent-27, score-0.473]

7 It has been suggested that deep architectures are more powerful in the sense of being able to more efﬁciently represent highly-varying functions [4, 3]. [sent-28, score-0.52]

8 The ﬁrst successful algorithms for training deep architectures appeared in 2006, with efﬁcient training procedures for Deep Belief Networks [14] and deep auto-encoders [13, 27, 6], both exploiting the general idea of greedy layer-wise pre-training [6]. [sent-31, score-0.88]

9 Poon and Domingos [25] introduced deep sum-product networks as a method to compute partition functions of tractable graphical models. [sent-33, score-0.548]

10 These networks are analogous to traditional artiﬁcial neural networks but with nodes that compute either products or weighted sums of their inputs. [sent-34, score-0.606]

11 If the nodes are organized in layers, we deﬁne the “hidden” layers to be those that are neither the input layer nor the output layer. [sent-36, score-0.461]

12 Whether such deep architectures have theoretical advantages compared to so-called “shallow” architectures (i. [sent-40, score-0.572]

13 The argument supported by our theoretical analysis is that a deep architecture is able to compute some functions much more efﬁciently than a shallow one. [sent-44, score-0.958]

14 Until recently, very few theoretical results supported the idea that deep architectures could present an advantage in terms of representing some functions more efﬁciently. [sent-45, score-0.52]

15 Another recent result on boolean circuits by Braverman [8] offers proof of a longstanding conjecture, showing that bounded-depth boolean circuits are unable to distinguish some (non-uniform) input distributions from the uniform distribution (i. [sent-50, score-0.455]

16 In particular, Braverman’s result suggests that shallow circuits can in general be fooled more easily than deep ones, i. [sent-53, score-0.995]

17 For instance, H˚ stad and Goldmann [12] extended some results to the case of networks of linear threshold units a with positivity constraints on the weights. [sent-58, score-0.498]

18 Bengio [3] suggests that some polynomials could be represented more efﬁciently by deep sumproduct networks, but without providing any formal statement or proofs. [sent-64, score-0.473]

19 This work partly addresses this void by demonstrating families of circuits for which a deep architecture can be exponentially more efﬁcient than a shallow one in the context of real-valued polynomials. [sent-65, score-1.125]

20 Note that we do not address in this paper the problem of learning these parameters: even if an efﬁcient deep representation exists for the function we seek to approximate, in general there is no 2 guarantee for standard optimization algorithms to easily converge to this representation. [sent-66, score-0.396]

21 This paper focuses on the representational power of deep sum-product circuits compared to shallow ones, and studies it by considering particular families of target functions (to be represented by the learner). [sent-67, score-1.108]

22 We consider two families of functions represented by deep sum-product networks (families F and G). [sent-69, score-0.625]

23 For each family, we establish a lower bound on the minimal number of hidden units a depth-2 sum-product network would require to represent a function of this family, showing it is much less efﬁcient than the deep representation. [sent-70, score-1.111]

24 A sum-product network is a network composed of units that either compute the product of their inputs or a weighted sum of their inputs (where weights are strictly positive). [sent-72, score-1.026]

25 Here, we restrict our deﬁnition of the generic term “sum-product network” to networks whose summation units have positive incoming weights1 , while others are called “negative-weight” networks. [sent-73, score-0.521]

26 A “negative-weight“ sum-product network may contain summation units whose weights are non-positive (i. [sent-75, score-0.617]

27 Finally, we formally deﬁne what we mean by deep vs. [sent-78, score-0.419]

28 A “shallow“ sum-product network contains a single hidden layer (i. [sent-81, score-0.495]

29 a total of three layers when counting the input and output layers, and a depth equal to two). [sent-83, score-0.45]

30 A “deep“ sum-product network contains more than one hidden layer (i. [sent-85, score-0.495]

31 1 Deﬁnition The ﬁrst family of functions we study, denoted by F, is made of functions built from deep sumproduct networks that alternate layers of product and sum units with two inputs each (details are provided below). [sent-89, score-1.381]

32 using a deep architecture) is equivalent to using a factorized representation of the polynomial function computed by the network. [sent-92, score-0.42]

33 Such a factorized representation can be exponentially more compact than its expansion as a sum of products (which can be associated to a shallow network with product units in its hidden layer and a sum unit as output). [sent-93, score-1.938]

34 2 Theoretical results The main result of this section is presented below in Corollary 1, providing a lower bound on the minimum number of hidden units required by a shallow sum-product network to represent a function f ∈ F. [sent-110, score-1.151]

35 (2) Show that the only possible architecture for a shallow sum-product network to compute f is to have a hidden layer made of product units, with a sum unit as output (Lemmas 2 to 5). [sent-112, score-1.389]

36 (3) Conclude that the number of hidden units must be at least the number of unique products computed in step 3. [sent-113, score-0.884]

37 Any element ℓk can be written as a (positively) weighted sum of products of input varij ables, such that each input variable xt is used in exactly one unit of ℓk . [sent-116, score-0.774]

38 Moreover, the number mk of products found in the sum computed by ℓk does not depend on j and obeys the following recurrence j rule for k ≥ 0: if k + 1 is odd, then mk+1 = m2 , otherwise mk+1 = 2mk . [sent-117, score-0.444]

39 By the inductive hypothesis, it is the product of two (positively) weighted sums of products of input variables, and no input variable can k appear in both ℓk 2j−1 and ℓ2j , so the result is also a (positively) weighted sum of products k of input variables. [sent-122, score-1.06]

40 Additionally, if the number of products in ℓk 2j−1 and ℓ2j is mk , then 2 mk+1 = mk , since all products involved in the multiplication of the two units are different (since they use disjoint subsets of input variables), and the sums have positive weights. [sent-123, score-1.131]

41 This unit is an input to a single unit of ℓk+1 , that will thus be the only unit of ℓk+1 where this input variable appears. [sent-125, score-0.595]

42 Again, from the induction assumption, it j must be a (positively) weighted sum of products of input variables, but with mk+1 = 2mk such products. [sent-127, score-0.599]

43 As in the previous case, an input variable will appear in the single unit of ℓk+1 that has as input the single unit of ℓk in which this variable must appear. [sent-128, score-0.563]

44 The number of products in the sum computed in the output unit l1 of a network √ n−1 . [sent-130, score-0.748]

45 This is obviously true for k = 1 since 2 = 2 = 1, and all units in ℓ1 are mk = 2 single products of the form xr xs . [sent-133, score-0.665]

46 The products computed in the output unit l1 can be split in two groups, one with products containing only variables x1 , . [sent-137, score-0.771]

47 This is obvious since the last unit is a “sum“ unit that adds two terms whose inputs are these two groups of variables (see e. [sent-145, score-0.371]

48 The products computed in the output unit l1 involve more than one input variable. [sent-150, score-0.57]

49 It is straightforward to show by induction on k ≥ 1 that the products computed by lj all involve more than one input variable, thus it is true in particular for the output layer (k = 2i). [sent-152, score-0.662]

50 Any shallow sum-product network computing f ∈ F must have a “sum” unit as output. [sent-154, score-0.827]

51 By contradiction, suppose the output unit of such a shallow sum-product network is multiplicative. [sent-156, score-0.863]

52 Thus the last unit must compute a product of two or more hidden units. [sent-158, score-0.439]

53 It can be re-written as a product of two factors, where each factor corresponds to either one hidden unit, or a product of multiple hidden units (it does not matter here which speciﬁc factorization is chosen among all possible ones). [sent-159, score-0.827]

54 Regardless of the type (sum or product) of the hidden units involved, those two factors can thus be written as weighted sums of products of variables xt (with positive weights, and input variables potentially raised to powers above one). [sent-160, score-1.118]

55 , xn then cannot be present in the second factor, since otherwise one product in the output 2 would contain both x1 and one of these variables (this product cannot cancel out since weights must be positive), violating Lemma 2. [sent-166, score-0.369]

56 Any shallow sum-product network computing f ∈ F must have only multiplicative units in its hidden layer. [sent-173, score-1.253]

57 By contradiction, suppose there exists a “sum“ unit in the hidden layer, written s = t∈S αt xt with S the set of input indices appearing in this sum, and αt > 0 for all t ∈ S. [sent-175, score-0.485]

58 Since according to Lemma 4 the output unit must also be a sum (and have positive weights according to Deﬁnition 1), then the ﬁnal output will also contain terms of the form βt xt for t ∈ S, with βt > 0. [sent-176, score-0.535]

59 Any shallow negative-weight sum-product network (see Deﬁnition 2) computing f ∈ F √ must have at least 2 n−1 hidden units, if its output unit is a sum and its hidden units are products. [sent-179, score-1.722]

60 Such a network computes a weighted sum of its hidden units, where each hidden unit is a γ product of input variables, i. [sent-181, score-0.959]

61 In order to compute a function in F, this shallow network thus needs a number of hidden units at least equal to the number of unique products in that function. [sent-184, score-1.472]

62 Any shallow sum-product network computing f ∈ F must have at least 2 units. [sent-187, score-0.719]

63 This is a direct corollary of Lemmas 4 (showing the output unit is a sum), 5 (showing that hidden units are products), and 6 (showing the desired result for any shallow network with this speciﬁc structure – regardless of the sign of weights). [sent-189, score-1.44]

64 3 Discussion Corollary 1 above shows that in order to compute some function in F with n inputs, the number of √ √ units in a shallow network has to be at least 2 n−1 , (i. [sent-191, score-0.992]

65 On another hand, the total number of units in the deep (for i > 1) network computing the same function, as described in Section 3. [sent-194, score-0.899]

66 It shows that some deep sumproduct network with n inputs and depth O(log n) can represent with O(n) units what would √ require O(2 n ) units for a depth-2 network. [sent-201, score-1.463]

67 Lemma 6 also shows a similar result regardless of the sign of the weights in the summation units of the depth-2 network, but assumes a speciﬁc architecture for this network (products in the hidden layer with a sum as output). [sent-202, score-1.079]

68 Compared to F, one important difference of deep sum-product networks built to deﬁne functions in G is that they can vary their input size independently of their depth. [sent-204, score-0.628]

69 Their analysis thus provides additional insight when comparing the representational efﬁciency of deep vs. [sent-205, score-0.434]

70 shallow sum-product networks in the case of a ﬁxed dataset. [sent-206, score-0.577]

71 1 Deﬁnition Networks in family G also alternate sum and product layers, but their units have as inputs all units from the previous layer except one. [sent-208, score-1.082]

72 More formally, deﬁne the family G = ∪n≥2,i≥0 Gin of functions represented by sum-product networks, where the sub-family Gin is made of all sum-product networks with n input variables and 2i + 2 layers (including the input layer ℓ0 ), such that: 1. [sent-209, score-0.677]

73 All layers are of size n, except the last layer ℓ2i+1 that contains a single sum unit that sums all units in the previous layer ℓ2i . [sent-214, score-0.971]

74 2 Theoretical results The main result is stated in Proposition 3 below, establishing a lower bound on the number of hidden units of a shallow sum-product network computing g ∈ G. [sent-225, score-1.151]

75 For any 0 ≤ j ≤ i, and any product of variables P = Πn xαt such that αt ∈ N and t=1 t j 2j whose computed value, when expanded as a weighted t αt = (n − 1) , there exists a unit in ℓ sum of products, contains P among these products. [sent-248, score-0.42]

76 As a result, by property 4 kq (in the deﬁnition of family G), each Pq will also appear in the additive layer ℓ2j+1 , in at least n − 1 different units (the only sum unit that may not contain Pq is the one that does not have ℓ2j as input). [sent-261, score-0.827]

77 kq By Lemma 7, we can thus ﬁnd a set of units ℓ2j+1 such that for any 1 ≤ q ≤ n − 1, the product rq Pq appears in ℓ2j+1 , with indices rq being different from each other. [sent-262, score-0.546]

78 Then, from property 4 of family G, the multiplicative unit ℓs computes the n−1 2j+1 product Πq=1 ℓrq , and as a result, when expanded as a sum of products, it contains in particular P1 . [sent-264, score-0.432]

79 The output gin of a sum-product network in Gin , when expanded as a sum of products, contains all products of variables of the form Πn xαt such that αt ∈ N and t αt = (n − 1)i . [sent-270, score-0.801]

80 Applying Proposition 2 with j = i, we obtain that all products of this form can be found in the multiplicative units of ℓ2i . [sent-272, score-0.617]

81 Since the output unit ℓ2i+1 computes a sum of these multiplicative 1 units (weighted with positive weights), those products are also present in the output. [sent-273, score-0.921]

82 A shallow negative-weight sum-product network computing gin ∈ Gin must have at least (n − 1)i hidden units. [sent-275, score-1.05]

83 First suppose the output unit of the shallow network is a sum. [sent-277, score-0.863]

84 Then it may be able to compute gin , assuming we allow multiplicative units in the hidden layer in the hidden layer to use powers of their inputs in the product they compute (which we allow here for the proof to be more generic). [sent-278, score-1.308]

85 However, it will require at least as many of these units as the number of unique products that can be found in the expansion of gin . [sent-279, score-0.794]

86 Then there can be no multiplicative hidden unit, otherwise it would mean one could factor some input variable xt in the computed function output: this is not possible since by Corollary 2, for any variable xt there exist products in the output function that do not involve xt . [sent-286, score-0.896]

87 So all hidden units must be additive, and since the computed function contains products of degree dni , there must be at least dni such hidden units. [sent-287, score-1.265]

88 3 Discussion Proposition 3 shows that in order to compute the same function as gin ∈ Gin , the number of units in the shallow network has to grow exponentially in i, i. [sent-289, score-1.157]

89 in the network’s depth (while the deep network’s size grows linearly in i). [sent-291, score-0.529]

90 The shallow network also needs to grow polynomially in the number of input variables n (with a degree equal to i), while the deep network grows only linearly in n. [sent-292, score-1.38]

91 It means that some deep sum-product network with n inputs and depth O(i) can represent with O(ni) units what would require O((n − 1)i ) units for a depth-2 network. [sent-293, score-1.417]

92 Note that in the similar results found for family F, the depth-2 network computing the same function as a function in F had to be constrained to either have a speciﬁc combination of sum and hidden units (in Lemma 6) or to have non-negative weights (in Corollary 1). [sent-294, score-0.876]

93 5 Conclusion We compared a deep sum-product network and a shallow sum-product network representing the same function, taken from two families of functions F and G. [sent-296, score-1.326]

94 For both families, we have shown that the number of units in the shallow network has to grow exponentially, compared to a linear growth in the deep network, so as to represent the same functions. [sent-297, score-1.383]

95 The deep version thus offers a much more compact representation of the same functions. [sent-298, score-0.396]

96 up to an error bound ǫ) with a much smaller shallow network. [sent-302, score-0.461]

97 An empirical evaluation of deep architectures on problems with many factors of variation. [sent-410, score-0.484]

98 Convolutional deep belief networks for scalable unsupervised learning of hierarchical representations. [sent-426, score-0.552]

99 Unsupervised feature learning for audio classiﬁcation using convolutional deep belief networks. [sent-434, score-0.466]

100 Semi-supervised learning of compact document representations with deep networks. [sent-473, score-0.396]

similar papers computed by tfidf model

tfidf for this paper:

wordName wordTfidf (topN-words)

[('shallow', 0.461), ('deep', 0.396), ('units', 0.325), ('products', 0.242), ('hidden', 0.187), ('network', 0.178), ('gin', 0.144), ('unit', 0.136), ('depth', 0.133), ('layer', 0.13), ('layers', 0.124), ('networks', 0.116), ('circuit', 0.1), ('circuits', 0.1), ('induction', 0.098), ('mk', 0.098), ('pq', 0.097), ('architectures', 0.088), ('output', 0.088), ('dni', 0.084), ('proposition', 0.082), ('input', 0.08), ('summation', 0.08), ('sum', 0.08), ('lemma', 0.079), ('families', 0.077), ('boolean', 0.075), ('hinton', 0.075), ('family', 0.072), ('corollary', 0.065), ('architecture', 0.065), ('product', 0.064), ('rq', 0.063), ('braverman', 0.062), ('inputs', 0.06), ('bengio', 0.059), ('xt', 0.057), ('stad', 0.057), ('jk', 0.057), ('must', 0.052), ('domingos', 0.05), ('multiplicative', 0.05), ('odd', 0.048), ('weighted', 0.047), ('sumproduct', 0.046), ('delalleau', 0.046), ('sums', 0.046), ('positively', 0.044), ('poon', 0.043), ('weston', 0.042), ('ranzato', 0.042), ('belief', 0.04), ('variables', 0.039), ('nodes', 0.039), ('fooled', 0.038), ('goldmann', 0.038), ('orponen', 0.038), ('stracuzzi', 0.038), ('utgoff', 0.038), ('representational', 0.038), ('collobert', 0.038), ('lecun', 0.038), ('functions', 0.036), ('weights', 0.034), ('predecessor', 0.033), ('polynomials', 0.031), ('montr', 0.031), ('kq', 0.031), ('powers', 0.031), ('expanded', 0.03), ('convolutional', 0.03), ('motivate', 0.029), ('expansion', 0.029), ('xn', 0.028), ('ng', 0.028), ('least', 0.028), ('parity', 0.027), ('variable', 0.027), ('qt', 0.027), ('nips', 0.026), ('exponentially', 0.026), ('salakhutdinov', 0.026), ('alternate', 0.026), ('pn', 0.026), ('contradiction', 0.026), ('mnih', 0.026), ('unique', 0.026), ('icml', 0.026), ('boltzmann', 0.025), ('showing', 0.025), ('equal', 0.025), ('pages', 0.025), ('written', 0.025), ('nition', 0.025), ('appear', 0.025), ('computed', 0.024), ('cortex', 0.024), ('mit', 0.024), ('grow', 0.023), ('formally', 0.023)]

similar papers list:

simIndex simValue paperId paperTitle

same-paper 1 0.99999976 250 nips-2011-Shallow vs. Deep Sum-Product Networks

Author: Olivier Delalleau, Yoshua Bengio

2 0.20175479 156 nips-2011-Learning to Learn with Compound HD Models

Author: Antonio Torralba, Joshua B. Tenenbaum, Ruslan Salakhutdinov

Abstract: We introduce HD (or “Hierarchical-Deep”) models, a new compositional learning architecture that integrates deep learning models with structured hierarchical Bayesian models. Speciﬁcally we show how we can learn a hierarchical Dirichlet process (HDP) prior over the activities of the top-level features in a Deep Boltzmann Machine (DBM). This compound HDP-DBM model learns to learn novel concepts from very few training examples, by learning low-level generic features, high-level features that capture correlations among low-level features, and a category hierarchy for sharing priors over the high-level features that are typical of different kinds of concepts. We present efﬁcient learning and inference algorithms for the HDP-DBM model and show that it is able to learn new concepts from very few examples on CIFAR-100 object recognition, handwritten character recognition, and human motion capture datasets. 1

3 0.1966684 244 nips-2011-Selecting Receptive Fields in Deep Networks

Author: Adam Coates, Andrew Y. Ng

Abstract: Recent deep learning and unsupervised feature learning systems that learn from unlabeled data have achieved high performance in benchmarks by using extremely large architectures with many features (hidden units) at each layer. Unfortunately, for such large architectures the number of parameters can grow quadratically in the width of the network, thus necessitating hand-coded “local receptive ﬁelds” that limit the number of connections from lower level features to higher ones (e.g., based on spatial locality). In this paper we propose a fast method to choose these connections that may be incorporated into a wide variety of unsupervised training methods. Speciﬁcally, we choose local receptive ﬁelds that group together those low-level features that are most similar to each other according to a pairwise similarity metric. This approach allows us to harness the advantages of local receptive ﬁelds (such as improved scalability, and reduced data requirements) when we do not know how to specify such receptive ﬁelds by hand or where our unsupervised training algorithm has no obvious generalization to a topographic setting. We produce results showing how this method allows us to use even simple unsupervised training algorithms to train successful multi-layered networks that achieve state-of-the-art results on CIFAR and STL datasets: 82.0% and 60.1% accuracy, respectively. 1

4 0.17883229 92 nips-2011-Expressive Power and Approximation Errors of Restricted Boltzmann Machines

Author: Guido F. Montufar, Johannes Rauh, Nihat Ay

Abstract: We present explicit classes of probability distributions that can be learned by Restricted Boltzmann Machines (RBMs) depending on the number of units that they contain, and which are representative for the expressive power of the model. We use this to show that the maximal Kullback-Leibler divergence to the RBM model with n visible and m hidden units is bounded from above by (n−1)−log(m+1). In this way we can specify the number of hidden units that guarantees a sufﬁciently rich model containing different classes of distributions and respecting a given error tolerance. 1

5 0.15124632 94 nips-2011-Facial Expression Transfer with Input-Output Temporal Restricted Boltzmann Machines

Author: Matthew D. Zeiler, Graham W. Taylor, Leonid Sigal, Iain Matthews, Rob Fergus

Abstract: We present a type of Temporal Restricted Boltzmann Machine that deﬁnes a probability distribution over an output sequence conditional on an input sequence. It shares the desirable properties of RBMs: efﬁcient exact inference, an exponentially more expressive latent state than HMMs, and the ability to model nonlinear structure and dynamics. We apply our model to a challenging real-world graphics problem: facial expression transfer. Our results demonstrate improved performance over several baselines modeling high-dimensional 2D and 3D data. 1

6 0.12451117 217 nips-2011-Practical Variational Inference for Neural Networks

7 0.11785769 287 nips-2011-The Manifold Tangent Classifier

8 0.11686397 276 nips-2011-Structured sparse coding via lateral inhibition

9 0.10442755 249 nips-2011-Sequence learning with hidden units in spiking neural networks

10 0.10319551 184 nips-2011-Neuronal Adaptation for Sampling-Based Probabilistic Inference in Perceptual Bistability

11 0.10179754 261 nips-2011-Sparse Filtering

12 0.08688727 93 nips-2011-Extracting Speaker-Specific Information with a Regularized Siamese Deep Network

13 0.086249784 197 nips-2011-On Tracking The Partition Function

14 0.084833525 113 nips-2011-Hierarchical Matching Pursuit for Image Classification: Architecture and Fast Algorithms

15 0.078795366 149 nips-2011-Learning Sparse Representations of High Dimensional Data on Large Scale Dictionaries

16 0.06984967 30 nips-2011-Algorithms for Hyper-Parameter Optimization

17 0.069163434 74 nips-2011-Dynamic Pooling and Unfolding Recursive Autoencoders for Paraphrase Detection

18 0.065324418 39 nips-2011-Approximating Semidefinite Programs in Sublinear Time

19 0.065302357 60 nips-2011-Confidence Sets for Network Structure

20 0.064878717 96 nips-2011-Fast and Balanced: Efficient Label Tree Learning for Large Scale Object Recognition

similar papers computed by lsi model

lsi for this paper:

topicId topicWeight

[(0, 0.193), (1, 0.058), (2, 0.026), (3, 0.002), (4, 0.014), (5, -0.08), (6, 0.09), (7, 0.082), (8, -0.058), (9, -0.308), (10, -0.029), (11, -0.166), (12, 0.053), (13, -0.053), (14, -0.031), (15, -0.072), (16, -0.012), (17, 0.033), (18, -0.092), (19, 0.105), (20, 0.101), (21, 0.07), (22, -0.17), (23, -0.032), (24, 0.0), (25, -0.071), (26, 0.097), (27, -0.134), (28, 0.032), (29, -0.022), (30, 0.091), (31, -0.0), (32, 0.078), (33, 0.099), (34, 0.176), (35, 0.141), (36, -0.005), (37, 0.07), (38, -0.074), (39, 0.024), (40, 0.011), (41, 0.025), (42, 0.105), (43, 0.042), (44, 0.027), (45, -0.044), (46, 0.108), (47, 0.061), (48, 0.061), (49, 0.003)]

similar papers list:

simIndex simValue paperId paperTitle

same-paper 1 0.97921765 250 nips-2011-Shallow vs. Deep Sum-Product Networks

Author: Olivier Delalleau, Yoshua Bengio

2 0.70937359 156 nips-2011-Learning to Learn with Compound HD Models

Author: Antonio Torralba, Joshua B. Tenenbaum, Ruslan Salakhutdinov

3 0.69831282 92 nips-2011-Expressive Power and Approximation Errors of Restricted Boltzmann Machines

Author: Guido F. Montufar, Johannes Rauh, Nihat Ay

4 0.69214022 93 nips-2011-Extracting Speaker-Specific Information with a Regularized Siamese Deep Network

Author: Ke Chen, Ahmad Salman

Abstract: Speech conveys different yet mixed information ranging from linguistic to speaker-speciﬁc components, and each of them should be exclusively used in a speciﬁc task. However, it is extremely difﬁcult to extract a speciﬁc information component given the fact that nearly all existing acoustic representations carry all types of speech information. Thus, the use of the same representation in both speech and speaker recognition hinders a system from producing better performance due to interference of irrelevant information. In this paper, we present a deep neural architecture to extract speaker-speciﬁc information from MFCCs. As a result, a multi-objective loss function is proposed for learning speaker-speciﬁc characteristics and regularization via normalizing interference of non-speaker related information and avoiding information loss. With LDC benchmark corpora and a Chinese speech corpus, we demonstrate that a resultant speaker-speciﬁc representation is insensitive to text/languages spoken and environmental mismatches and hence outperforms MFCCs and other state-of-the-art techniques in speaker recognition. We discuss relevant issues and relate our approach to previous work. 1

5 0.61862379 94 nips-2011-Facial Expression Transfer with Input-Output Temporal Restricted Boltzmann Machines

Author: Matthew D. Zeiler, Graham W. Taylor, Leonid Sigal, Iain Matthews, Rob Fergus

6 0.6131565 184 nips-2011-Neuronal Adaptation for Sampling-Based Probabilistic Inference in Perceptual Bistability

7 0.53151733 74 nips-2011-Dynamic Pooling and Unfolding Recursive Autoencoders for Paraphrase Detection

8 0.50913334 244 nips-2011-Selecting Receptive Fields in Deep Networks

9 0.48305631 197 nips-2011-On Tracking The Partition Function

10 0.47907615 287 nips-2011-The Manifold Tangent Classifier

11 0.45949829 217 nips-2011-Practical Variational Inference for Neural Networks

12 0.45900843 249 nips-2011-Sequence learning with hidden units in spiking neural networks

13 0.41086251 60 nips-2011-Confidence Sets for Network Structure

14 0.39902338 276 nips-2011-Structured sparse coding via lateral inhibition

15 0.35373056 6 nips-2011-A Global Structural EM Algorithm for a Model of Cancer Progression

16 0.34355351 30 nips-2011-Algorithms for Hyper-Parameter Optimization

17 0.33551884 243 nips-2011-Select and Sample - A Model of Efficient Neural Inference and Learning

18 0.30856583 113 nips-2011-Hierarchical Matching Pursuit for Image Classification: Architecture and Fast Algorithms

19 0.2968151 115 nips-2011-Hierarchical Topic Modeling for Analysis of Time-Evolving Personal Choices

20 0.29623911 292 nips-2011-Two is better than one: distinct roles for familiarity and recollection in retrieving palimpsest memories

similar papers computed by lda model

lda for this paper:

topicId topicWeight

[(0, 0.396), (4, 0.03), (20, 0.025), (26, 0.014), (31, 0.146), (43, 0.054), (45, 0.061), (57, 0.043), (65, 0.04), (74, 0.046), (83, 0.038), (99, 0.014)]

similar papers list:

simIndex simValue paperId paperTitle

1 0.9173671 207 nips-2011-Optimal learning rates for least squares SVMs using Gaussian kernels

Author: Mona Eberts, Ingo Steinwart

Abstract: We prove a new oracle inequality for support vector machines with Gaussian RBF kernels solving the regularized least squares regression problem. To this end, we apply the modulus of smoothness. With the help of the new oracle inequality we then derive learning rates that can also be achieved by a simple data-dependent parameter selection method. Finally, it turns out that our learning rates are asymptotically optimal for regression functions satisfying certain standard smoothness conditions. 1

same-paper 2 0.89973211 250 nips-2011-Shallow vs. Deep Sum-Product Networks

Author: Olivier Delalleau, Yoshua Bengio

3 0.84092414 211 nips-2011-Penalty Decomposition Methods for Rank Minimization

Author: Yong Zhang, Zhaosong Lu

Abstract: In this paper we consider general rank minimization problems with rank appearing in either objective function or constraint. We ﬁrst show that a class of matrix optimization problems can be solved as lower dimensional vector optimization problems. As a consequence, we establish that a class of rank minimization problems have closed form solutions. Using this result, we then propose penalty decomposition methods for general rank minimization problems. The convergence results of the PD methods have been shown in the longer version of the paper [19]. Finally, we test the performance of our methods by applying them to matrix completion and nearest low-rank correlation matrix problems. The computational results demonstrate that our methods generally outperform the existing methods in terms of solution quality and/or speed. 1

4 0.79330444 12 nips-2011-A Two-Stage Weighting Framework for Multi-Source Domain Adaptation

Author: Qian Sun, Rita Chattopadhyay, Sethuraman Panchanathan, Jieping Ye

Abstract: Discriminative learning when training and test data belong to different distributions is a challenging and complex task. Often times we have very few or no labeled data from the test or target distribution but may have plenty of labeled data from multiple related sources with different distributions. The difference in distributions may be both in marginal and conditional probabilities. Most of the existing domain adaptation work focuses on the marginal probability distribution difference between the domains, assuming that the conditional probabilities are similar. However in many real world applications, conditional probability distribution differences are as commonplace as marginal probability differences. In this paper we propose a two-stage domain adaptation methodology which combines weighted data from multiple sources based on marginal probability differences (ﬁrst stage) as well as conditional probability differences (second stage), with the target domain data. The weights for minimizing the marginal probability differences are estimated independently, while the weights for minimizing conditional probability differences are computed simultaneously by exploiting the potential interaction among multiple sources. We also provide a theoretical analysis on the generalization performance of the proposed multi-source domain adaptation formulation using the weighted Rademacher complexity measure. Empirical comparisons with existing state-of-the-art domain adaptation methods using three real-world datasets demonstrate the effectiveness of the proposed approach. 1

5 0.65253687 96 nips-2011-Fast and Balanced: Efficient Label Tree Learning for Large Scale Object Recognition

Author: Jia Deng, Sanjeev Satheesh, Alexander C. Berg, Fei Li

Abstract: We present a novel approach to efﬁciently learn a label tree for large scale classiﬁcation with many classes. The key contribution of the approach is a technique to simultaneously determine the structure of the tree and learn the classiﬁers for each node in the tree. This approach also allows ﬁne grained control over the efﬁciency vs accuracy trade-off in designing a label tree, leading to more balanced trees. Experiments are performed on large scale image classiﬁcation with 10184 classes and 9 million images. We demonstrate signiﬁcant improvements in test accuracy and efﬁciency with less training time and more balanced trees compared to the previous state of the art by Bengio et al. 1

6 0.61461431 152 nips-2011-Learning in Hilbert vs. Banach Spaces: A Measure Embedding Viewpoint

7 0.57890731 256 nips-2011-Solving Decision Problems with Limited Information

8 0.55126297 92 nips-2011-Expressive Power and Approximation Errors of Restricted Boltzmann Machines

9 0.54561669 106 nips-2011-Generalizing from Several Related Classification Tasks to a New Unlabeled Sample

10 0.53674394 53 nips-2011-Co-Training for Domain Adaptation

11 0.52817279 297 nips-2011-Universal low-rank matrix recovery from Pauli measurements

12 0.51974887 265 nips-2011-Sparse recovery by thresholded non-negative least squares

13 0.51887715 300 nips-2011-Variance Reduction in Monte-Carlo Tree Search

14 0.51299566 197 nips-2011-On Tracking The Partition Function

15 0.5129475 116 nips-2011-Hierarchically Supervised Latent Dirichlet Allocation

16 0.51242948 237 nips-2011-Reinforcement Learning using Kernel-Based Stochastic Factorization

17 0.51225114 47 nips-2011-Beyond Spectral Clustering - Tight Relaxations of Balanced Graph Cuts

18 0.51111501 156 nips-2011-Learning to Learn with Compound HD Models

19 0.50988311 291 nips-2011-Transfer from Multiple MDPs

20 0.5097338 301 nips-2011-Variational Gaussian Process Dynamical Systems