Complex information processing systems that are capable of a wide variety of tasks, such as the human brain, are composed of specialized units that collaborate and communicate with each other. An important property of such information processing networks is locality: there is no single global unit controlling the modules, but information is exchanged locally. Here, we consider a decision-theoretic approach to study networks of bounded rational decision makers that are allowed to specialize and communicate with each other. In contrast to previous work that has focused on feedforward communication between decision-making agents, we consider cyclical information processing paths allowing for back-and-forth communication. We adapt message-passing algorithms to suit this purpose, essentially allowing for local information flow between units and thus enabling circular dependency structures. We provide examples that show how repeated communication can increase performance given that each unit’s information processing capability is limited and that decision-making systems with too few or too many connections and feedback loops achieve suboptimal utility.
{"title":"Bounded Rational Decision Networks With Belief Propagation","authors":"Gerrit Schmid;Sebastian Gottwald;Daniel A. Braun","doi":"10.1162/neco_a_01719","DOIUrl":"10.1162/neco_a_01719","url":null,"abstract":"Complex information processing systems that are capable of a wide variety of tasks, such as the human brain, are composed of specialized units that collaborate and communicate with each other. An important property of such information processing networks is locality: there is no single global unit controlling the modules, but information is exchanged locally. Here, we consider a decision-theoretic approach to study networks of bounded rational decision makers that are allowed to specialize and communicate with each other. In contrast to previous work that has focused on feedforward communication between decision-making agents, we consider cyclical information processing paths allowing for back-and-forth communication. We adapt message-passing algorithms to suit this purpose, essentially allowing for local information flow between units and thus enabling circular dependency structures. We provide examples that show how repeated communication can increase performance given that each unit’s information processing capability is limited and that decision-making systems with too few or too many connections and feedback loops achieve suboptimal utility.","PeriodicalId":54731,"journal":{"name":"Neural Computation","volume":"37 1","pages":"76-127"},"PeriodicalIF":2.7,"publicationDate":"2024-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10810330","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142395372","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Max Dabagia;Christos H. Papadimitriou;Santosh S. Vempala
Even as machine learning exceeds human-level performance on many applications, the generality, robustness, and rapidity of the brain’s learning capabilities remain unmatched. How cognition arises from neural activity is the central open question in neuroscience, inextricable from the study of intelligence itself. A simple formal model of neural activity was proposed in Papadimitriou et al. (2020) and has been subsequently shown, through both mathematical proofs and simulations, to be capable of implementing certain simple cognitive operations via the creation and manipulation of assemblies of neurons. However, many intelligent behaviors rely on the ability to recognize, store, and manipulate temporal sequences of stimuli (planning, language, navigation, to list a few). Here we show that in the same model, sequential precedence can be captured naturally through synaptic weights and plasticity, and, as a result, a range of computations on sequences of assemblies can be carried out. In particular, repeated presentation of a sequence of stimuli leads to the memorization of the sequence through corresponding neural assemblies: upon future presentation of any stimulus in the sequence, the corresponding assembly and its subsequent ones will be activated, one after the other, until the end of the sequence. If the stimulus sequence is presented to two brain areas simultaneously, a scaffolded representation is created, resulting in more efficient memorization and recall, in agreement with cognitive experiments. Finally, we show that any finite state machine can be learned in a similar way, through the presentation of appropriate patterns of sequences. Through an extension of this mechanism, the model can be shown to be capable of universal computation. Taken together, these results provide a concrete hypothesis for the basis of the brain’s remarkable abilities to compute and learn, with sequences playing a vital role.
{"title":"Computation With Sequences of Assemblies in a Model of the Brain","authors":"Max Dabagia;Christos H. Papadimitriou;Santosh S. Vempala","doi":"10.1162/neco_a_01720","DOIUrl":"10.1162/neco_a_01720","url":null,"abstract":"Even as machine learning exceeds human-level performance on many applications, the generality, robustness, and rapidity of the brain’s learning capabilities remain unmatched. How cognition arises from neural activity is the central open question in neuroscience, inextricable from the study of intelligence itself. A simple formal model of neural activity was proposed in Papadimitriou et al. (2020) and has been subsequently shown, through both mathematical proofs and simulations, to be capable of implementing certain simple cognitive operations via the creation and manipulation of assemblies of neurons. However, many intelligent behaviors rely on the ability to recognize, store, and manipulate temporal sequences of stimuli (planning, language, navigation, to list a few). Here we show that in the same model, sequential precedence can be captured naturally through synaptic weights and plasticity, and, as a result, a range of computations on sequences of assemblies can be carried out. In particular, repeated presentation of a sequence of stimuli leads to the memorization of the sequence through corresponding neural assemblies: upon future presentation of any stimulus in the sequence, the corresponding assembly and its subsequent ones will be activated, one after the other, until the end of the sequence. If the stimulus sequence is presented to two brain areas simultaneously, a scaffolded representation is created, resulting in more efficient memorization and recall, in agreement with cognitive experiments. Finally, we show that any finite state machine can be learned in a similar way, through the presentation of appropriate patterns of sequences. Through an extension of this mechanism, the model can be shown to be capable of universal computation. Taken together, these results provide a concrete hypothesis for the basis of the brain’s remarkable abilities to compute and learn, with sequences playing a vital role.","PeriodicalId":54731,"journal":{"name":"Neural Computation","volume":"37 1","pages":"193-233"},"PeriodicalIF":2.7,"publicationDate":"2024-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142395373","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Christopher J. Kymn;Denis Kleyko;E. Paxon Frady;Connor Bybee;Pentti Kanerva;Friedrich T. Sommer;Bruno A. Olshausen
We introduce residue hyperdimensional computing, a computing framework that unifies residue number systems with an algebra defined over random, high-dimensional vectors. We show how residue numbers can be represented as high-dimensional vectors in a manner that allows algebraic operations to be performed with component-wise, parallelizable operations on the vector elements. The resulting framework, when combined with an efficient method for factorizing high-dimensional vectors, can represent and operate on numerical values over a large dynamic range using resources that scale only logarithmically with the range, a vast improvement over previous methods. It also exhibits impressive robustness to noise. We demonstrate the potential for this framework to solve computationally difficult problems in visual perception and combinatorial optimization, showing improvement over baseline methods. More broadly, the framework provides a possible account for the computational operations of grid cells in the brain, and it suggests new machine learning architectures for representing and manipulating numerical data.
{"title":"Computing With Residue Numbers in High-Dimensional Representation","authors":"Christopher J. Kymn;Denis Kleyko;E. Paxon Frady;Connor Bybee;Pentti Kanerva;Friedrich T. Sommer;Bruno A. Olshausen","doi":"10.1162/neco_a_01723","DOIUrl":"10.1162/neco_a_01723","url":null,"abstract":"We introduce residue hyperdimensional computing, a computing framework that unifies residue number systems with an algebra defined over random, high-dimensional vectors. We show how residue numbers can be represented as high-dimensional vectors in a manner that allows algebraic operations to be performed with component-wise, parallelizable operations on the vector elements. The resulting framework, when combined with an efficient method for factorizing high-dimensional vectors, can represent and operate on numerical values over a large dynamic range using resources that scale only logarithmically with the range, a vast improvement over previous methods. It also exhibits impressive robustness to noise. We demonstrate the potential for this framework to solve computationally difficult problems in visual perception and combinatorial optimization, showing improvement over baseline methods. More broadly, the framework provides a possible account for the computational operations of grid cells in the brain, and it suggests new machine learning architectures for representing and manipulating numerical data.","PeriodicalId":54731,"journal":{"name":"Neural Computation","volume":"37 1","pages":"1-37"},"PeriodicalIF":2.7,"publicationDate":"2024-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142669937","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Tomohiro Shiraishi;Daiki Miwa;Vo Nguyen Le Duy;Ichiro Takeuchi
In this study, we investigate the quantification of the statistical reliability of detected change points (CPs) in time series using a recurrent neural network (RNN). Thanks to its flexibility, RNN holds the potential to effectively identify CPs in time series characterized by complex dynamics. However, there is an increased risk of erroneously detecting random noise fluctuations as CPs. The primary goal of this study is to rigorously control the risk of false detections by providing theoretically valid p-values to the CPs detected by RNN. To achieve this, we introduce a novel method based on the framework of selective inference (SI). SI enables valid inferences by conditioning on the event of hypothesis selection, thus mitigating bias from generating and testing hypotheses on the same data. In this study, we apply an SI framework to RNN-based CP detection, where characterizing the complex process of RNN selecting CPs is our main technical challenge. We demonstrate the validity and effectiveness of the proposed method through artificial and real data experiments.
{"title":"Selective Inference for Change Point Detection by Recurrent Neural Network","authors":"Tomohiro Shiraishi;Daiki Miwa;Vo Nguyen Le Duy;Ichiro Takeuchi","doi":"10.1162/neco_a_01724","DOIUrl":"10.1162/neco_a_01724","url":null,"abstract":"In this study, we investigate the quantification of the statistical reliability of detected change points (CPs) in time series using a recurrent neural network (RNN). Thanks to its flexibility, RNN holds the potential to effectively identify CPs in time series characterized by complex dynamics. However, there is an increased risk of erroneously detecting random noise fluctuations as CPs. The primary goal of this study is to rigorously control the risk of false detections by providing theoretically valid p-values to the CPs detected by RNN. To achieve this, we introduce a novel method based on the framework of selective inference (SI). SI enables valid inferences by conditioning on the event of hypothesis selection, thus mitigating bias from generating and testing hypotheses on the same data. In this study, we apply an SI framework to RNN-based CP detection, where characterizing the complex process of RNN selecting CPs is our main technical challenge. We demonstrate the validity and effectiveness of the proposed method through artificial and real data experiments.","PeriodicalId":54731,"journal":{"name":"Neural Computation","volume":"37 1","pages":"160-192"},"PeriodicalIF":2.7,"publicationDate":"2024-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142666703","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Michele Garibbo;Casimir J. H. Ludwig;Nathan F. Lepora;Laurence Aitchison
In human error–based learning, the size and direction of a scalar error (i.e., the “directed error”) are used to update future actions. Modern deep reinforcement learning (RL) methods perform a similar operation but in terms of scalar rewards. Despite this similarity, the relationship between action updates of deep RL and human error–based learning has yet to be investigated. Here, we systematically compare the three major families of deep RL algorithms to human error–based learning. We show that all three deep RL approaches are qualitatively different from human error–based learning, as assessed by a mirror-reversal perturbation experiment. To bridge this gap, we developed an alternative deep RL algorithm inspired by human error–based learning, model-based deterministic policy gradients (MB-DPG). We showed that MB-DPG captures human error–based learning under mirror-reversal and rotational perturbations and that MB-DPG learns faster than canonical model-free algorithms on complex arm-based reaching tasks, while being more robust to (forward) model misspecification than model-based RL.
{"title":"Relating Human Error–Based Learning to Modern Deep RL Algorithms","authors":"Michele Garibbo;Casimir J. H. Ludwig;Nathan F. Lepora;Laurence Aitchison","doi":"10.1162/neco_a_01721","DOIUrl":"10.1162/neco_a_01721","url":null,"abstract":"In human error–based learning, the size and direction of a scalar error (i.e., the “directed error”) are used to update future actions. Modern deep reinforcement learning (RL) methods perform a similar operation but in terms of scalar rewards. Despite this similarity, the relationship between action updates of deep RL and human error–based learning has yet to be investigated. Here, we systematically compare the three major families of deep RL algorithms to human error–based learning. We show that all three deep RL approaches are qualitatively different from human error–based learning, as assessed by a mirror-reversal perturbation experiment. To bridge this gap, we developed an alternative deep RL algorithm inspired by human error–based learning, model-based deterministic policy gradients (MB-DPG). We showed that MB-DPG captures human error–based learning under mirror-reversal and rotational perturbations and that MB-DPG learns faster than canonical model-free algorithms on complex arm-based reaching tasks, while being more robust to (forward) model misspecification than model-based RL.","PeriodicalId":54731,"journal":{"name":"Neural Computation","volume":"37 1","pages":"128-159"},"PeriodicalIF":2.7,"publicationDate":"2024-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142395376","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The free energy principle (FEP) describes (biological) agents as minimizing a variational free energy (FE) with respect to a generative model of their environment. Active inference (AIF) is a corollary of the FEP that describes how agents explore and exploit their environment by minimizing an expected FE objective. In two related papers, we describe a scalable, epistemic approach to synthetic AIF by message passing on free-form Forney-style factor graphs (FFGs). A companion paper (part I of this article; Koudahl et al., 2023) introduces a constrained FFG (CFFG) notation that visually represents (generalized) FE objectives for AIF. This article (part II) derives message-passing algorithms that minimize (generalized) FE objectives on a CFFG by variational calculus. A comparison between simulated Bethe and generalized FE agents illustrates how the message-passing approach to synthetic AIF induces epistemic behavior on a T-maze navigation task. Extension of the T-maze simulation to learning goal statistics and a multiagent bargaining setting illustrate how this approach encourages reuse of nodes and updates in alternative settings. With a full message-passing account of synthetic AIF agents, it becomes possible to derive and reuse message updates across models and move closer to industrial applications of synthetic AIF.
自由能原理（FEP）将（生物）代理描述为相对于其环境的生成模型最小化可变自由能（FE）。主动推理（AIF）是自由能原理的必然结果，它描述了生物体如何通过最小化预期自由能目标来探索和利用其环境。在两篇相关论文中，我们描述了通过在自由形式的福尼式因子图（FFGs）上进行消息传递来合成 AIF 的可扩展认识论方法。另一篇相关论文（本文第一部分；Koudahl 等人，2023 年）介绍了一种受限 FFG（CFFG）符号，它能直观地表示 AIF 的（广义）FE 目标。本文（第二部分）通过变分法推导了在 CFFG 上最小化（广义）FE 目标的消息传递算法。模拟贝特代理和广义 FE 代理之间的比较说明了合成 AIF 的信息传递方法如何在 T 型迷宫导航任务中诱导认识行为。将 T 型迷宫模拟扩展到学习目标统计和多代理讨价还价设置，说明了这种方法如何鼓励在其他设置中重复使用节点和更新。有了合成 AIF 代理的完整消息传递账户，就有可能在不同模型中推导和重用消息更新，并更接近合成 AIF 的工业应用。
{"title":"Realizing Synthetic Active Inference Agents, Part II: Variational Message Updates","authors":"Thijs van de Laar;Magnus Koudahl;Bert de Vries","doi":"10.1162/neco_a_01713","DOIUrl":"10.1162/neco_a_01713","url":null,"abstract":"The free energy principle (FEP) describes (biological) agents as minimizing a variational free energy (FE) with respect to a generative model of their environment. Active inference (AIF) is a corollary of the FEP that describes how agents explore and exploit their environment by minimizing an expected FE objective. In two related papers, we describe a scalable, epistemic approach to synthetic AIF by message passing on free-form Forney-style factor graphs (FFGs). A companion paper (part I of this article; Koudahl et al., 2023) introduces a constrained FFG (CFFG) notation that visually represents (generalized) FE objectives for AIF. This article (part II) derives message-passing algorithms that minimize (generalized) FE objectives on a CFFG by variational calculus. A comparison between simulated Bethe and generalized FE agents illustrates how the message-passing approach to synthetic AIF induces epistemic behavior on a T-maze navigation task. Extension of the T-maze simulation to learning goal statistics and a multiagent bargaining setting illustrate how this approach encourages reuse of nodes and updates in alternative settings. With a full message-passing account of synthetic AIF agents, it becomes possible to derive and reuse message updates across models and move closer to industrial applications of synthetic AIF.","PeriodicalId":54731,"journal":{"name":"Neural Computation","volume":"37 1","pages":"38-75"},"PeriodicalIF":2.7,"publicationDate":"2024-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142309093","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Given a directed graph of nodes and edges connecting them, a common problem is to find the shortest path between any two nodes. Here we show that the shortest path distances can be found by a simple matrix inversion: if the edges are given by the adjacency matrix Aij, then with a suitably small value of γ, the shortest path distances are Dij=ceil(logγ[(I-γA)-1]ij).We derive several graph-theoretic bounds on the value of γ and explore its useful range with numerics on different graph types. Even when the distance function is not globally accurate across the entire graph, it still works locally to instruct pursuit of the shortest path. In this mode, it also extends to weighted graphs with positive edge weights. For a wide range of dense graphs, this distance function is computationally faster than the best available alternative. Finally, we show that this method leads naturally to a neural network solution of the all-pairs-shortest-path problem.
{"title":"A Fast Algorithm for All-Pairs-Shortest-Paths Suitable for Neural Networks","authors":"Zeyu Jing;Markus Meister","doi":"10.1162/neco_a_01716","DOIUrl":"10.1162/neco_a_01716","url":null,"abstract":"Given a directed graph of nodes and edges connecting them, a common problem is to find the shortest path between any two nodes. Here we show that the shortest path distances can be found by a simple matrix inversion: if the edges are given by the adjacency matrix Aij, then with a suitably small value of γ, the shortest path distances are Dij=ceil(logγ[(I-γA)-1]ij).We derive several graph-theoretic bounds on the value of γ and explore its useful range with numerics on different graph types. Even when the distance function is not globally accurate across the entire graph, it still works locally to instruct pursuit of the shortest path. In this mode, it also extends to weighted graphs with positive edge weights. For a wide range of dense graphs, this distance function is computationally faster than the best available alternative. Finally, we show that this method leads naturally to a neural network solution of the all-pairs-shortest-path problem.","PeriodicalId":54731,"journal":{"name":"Neural Computation","volume":"36 12","pages":"2710-2733"},"PeriodicalIF":2.7,"publicationDate":"2024-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142395362","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Vasily Zadorozhnyy;Edison Mucllari;Cole Pospisil;Duc Nguyen;Qiang Ye
In recent years, using orthogonal matrices has been shown to be a promising approach to improving recurrent neural networks (RNNs) with training, stability, and convergence, particularly to control gradients. While gated recurrent unit (GRU) and long short-term memory (LSTM) architectures address the vanishing gradient problem by using a variety of gates and memory cells, they are still prone to the exploding gradient problem. In this work, we analyze the gradients in GRU and propose the use of orthogonal matrices to prevent exploding gradient problems and enhance long-term memory. We study where to use orthogonal matrices and propose a Neumann series–based scaled Cayley transformation for training orthogonal matrices in GRU, which we call Neumann-Cayley orthogonal GRU (NC-GRU). We present detailed experiments of our model on several synthetic and real-world tasks, which show that NC-GRU significantly outperforms GRU and several other RNNs.
{"title":"Orthogonal Gated Recurrent Unit With Neumann-Cayley Transformation","authors":"Vasily Zadorozhnyy;Edison Mucllari;Cole Pospisil;Duc Nguyen;Qiang Ye","doi":"10.1162/neco_a_01710","DOIUrl":"10.1162/neco_a_01710","url":null,"abstract":"In recent years, using orthogonal matrices has been shown to be a promising approach to improving recurrent neural networks (RNNs) with training, stability, and convergence, particularly to control gradients. While gated recurrent unit (GRU) and long short-term memory (LSTM) architectures address the vanishing gradient problem by using a variety of gates and memory cells, they are still prone to the exploding gradient problem. In this work, we analyze the gradients in GRU and propose the use of orthogonal matrices to prevent exploding gradient problems and enhance long-term memory. We study where to use orthogonal matrices and propose a Neumann series–based scaled Cayley transformation for training orthogonal matrices in GRU, which we call Neumann-Cayley orthogonal GRU (NC-GRU). We present detailed experiments of our model on several synthetic and real-world tasks, which show that NC-GRU significantly outperforms GRU and several other RNNs.","PeriodicalId":54731,"journal":{"name":"Neural Computation","volume":"36 12","pages":"2651-2676"},"PeriodicalIF":2.7,"publicationDate":"2024-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142309092","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Victor Geadah;Gabriel Barello;Daniel Greenidge;Adam S. Charles;Jonathan W. Pillow
The sparse coding model posits that the visual system has evolved to efficiently code natural stimuli using a sparse set of features from an overcomplete dictionary. The original sparse coding model suffered from two key limitations; however: (1) computing the neural response to an image patch required minimizing a nonlinear objective function via recurrent dynamics and (2) fitting relied on approximate inference methods that ignored uncertainty. Although subsequent work has developed several methods to overcome these obstacles, we propose a novel solution inspired by the variational autoencoder (VAE) framework. We introduce the sparse coding variational autoencoder (SVAE), which augments the sparse coding model with a probabilistic recognition model parameterized by a deep neural network. This recognition model provides a neurally plausible feedforward implementation for the mapping from image patches to neural activities and enables a principled method for fitting the sparse coding model to data via maximization of the evidence lower bound (ELBO). The SVAE differs from standard VAEs in three key respects: the latent representation is overcomplete (there are more latent dimensions than image pixels), the prior is sparse or heavy-tailed instead of gaussian, and the decoder network is a linear projection instead of a deep network. We fit the SVAE to natural image data under different assumed prior distributions and show that it obtains higher test performance than previous fitting methods. Finally, we examine the response properties of the recognition network and show that it captures important nonlinear properties of neurons in the early visual pathway.
{"title":"Sparse-Coding Variational Autoencoders","authors":"Victor Geadah;Gabriel Barello;Daniel Greenidge;Adam S. Charles;Jonathan W. Pillow","doi":"10.1162/neco_a_01715","DOIUrl":"10.1162/neco_a_01715","url":null,"abstract":"The sparse coding model posits that the visual system has evolved to efficiently code natural stimuli using a sparse set of features from an overcomplete dictionary. The original sparse coding model suffered from two key limitations; however: (1) computing the neural response to an image patch required minimizing a nonlinear objective function via recurrent dynamics and (2) fitting relied on approximate inference methods that ignored uncertainty. Although subsequent work has developed several methods to overcome these obstacles, we propose a novel solution inspired by the variational autoencoder (VAE) framework. We introduce the sparse coding variational autoencoder (SVAE), which augments the sparse coding model with a probabilistic recognition model parameterized by a deep neural network. This recognition model provides a neurally plausible feedforward implementation for the mapping from image patches to neural activities and enables a principled method for fitting the sparse coding model to data via maximization of the evidence lower bound (ELBO). The SVAE differs from standard VAEs in three key respects: the latent representation is overcomplete (there are more latent dimensions than image pixels), the prior is sparse or heavy-tailed instead of gaussian, and the decoder network is a linear projection instead of a deep network. We fit the SVAE to natural image data under different assumed prior distributions and show that it obtains higher test performance than previous fitting methods. Finally, we examine the response properties of the recognition network and show that it captures important nonlinear properties of neurons in the early visual pathway.","PeriodicalId":54731,"journal":{"name":"Neural Computation","volume":"36 12","pages":"2571-2601"},"PeriodicalIF":2.7,"publicationDate":"2024-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142395377","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Neural network pruning is a popular approach to reducing the computational costs of training and/or deploying a network and aims to do so while minimizing accuracy loss. Pruning methods that remove individual weights (fine granularity) can remove more total network parameters before reaching a given degree of accuracy loss, while methods that preserve some or all of a network’s structure (coarser granularity, such as pruning channels from a CNN) take better advantage of hardware and software optimized for dense matrix computations. We compare intelligent iterative pruning using several different criteria sampled from the literature against random pruning at initialization across multiple granularities on two different architectures and three image classification tasks. Our work is the first direct and comprehensive investigation of the relationship between granularity and the efficacy of intelligent pruning relative to a random-pruning baseline. We find that the accuracy advantage of intelligent over random pruning decreases dramatically as granularity becomes coarser, with minimal advantage for intelligent pruning at granularity coarse enough to fully preserve network structure. For instance, at pruning rates where random pruning leaves ResNet-20 at 85.0% test accuracy on CIFAR-10 after 30,000 training iterations, intelligent weight pruning with the best-in-context criterion leaves it at about 90.0% accuracy (on par with the unpruned network), kernel pruning leaves it at about 86.5%, and channel pruning leaves it at about 85.5%. Our results suggest that compared to coarse pruning, fine pruning combined with efficient implementation of the resulting networks is a more promising direction for easing the trade-off between high accuracy and low computational cost.
{"title":"Fine Granularity Is Critical for Intelligent Neural Network Pruning","authors":"Alex Heyman;Joel Zylberberg","doi":"10.1162/neco_a_01717","DOIUrl":"10.1162/neco_a_01717","url":null,"abstract":"Neural network pruning is a popular approach to reducing the computational costs of training and/or deploying a network and aims to do so while minimizing accuracy loss. Pruning methods that remove individual weights (fine granularity) can remove more total network parameters before reaching a given degree of accuracy loss, while methods that preserve some or all of a network’s structure (coarser granularity, such as pruning channels from a CNN) take better advantage of hardware and software optimized for dense matrix computations. We compare intelligent iterative pruning using several different criteria sampled from the literature against random pruning at initialization across multiple granularities on two different architectures and three image classification tasks. Our work is the first direct and comprehensive investigation of the relationship between granularity and the efficacy of intelligent pruning relative to a random-pruning baseline. We find that the accuracy advantage of intelligent over random pruning decreases dramatically as granularity becomes coarser, with minimal advantage for intelligent pruning at granularity coarse enough to fully preserve network structure. For instance, at pruning rates where random pruning leaves ResNet-20 at 85.0% test accuracy on CIFAR-10 after 30,000 training iterations, intelligent weight pruning with the best-in-context criterion leaves it at about 90.0% accuracy (on par with the unpruned network), kernel pruning leaves it at about 86.5%, and channel pruning leaves it at about 85.5%. Our results suggest that compared to coarse pruning, fine pruning combined with efficient implementation of the resulting networks is a more promising direction for easing the trade-off between high accuracy and low computational cost.","PeriodicalId":54731,"journal":{"name":"Neural Computation","volume":"36 12","pages":"2677-2709"},"PeriodicalIF":2.7,"publicationDate":"2024-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142395374","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：