1
|
Stöckl C, Yang Y, Maass W. Local prediction-learning in high-dimensional spaces enables neural networks to plan. Nat Commun 2024; 15:2344. [PMID: 38490999 PMCID: PMC10943103 DOI: 10.1038/s41467-024-46586-0] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 07/25/2023] [Accepted: 03/01/2024] [Indexed: 03/18/2024] Open
Abstract
Planning and problem solving are cornerstones of higher brain function. But we do not know how the brain does that. We show that learning of a suitable cognitive map of the problem space suffices. Furthermore, this can be reduced to learning to predict the next observation through local synaptic plasticity. Importantly, the resulting cognitive map encodes relations between actions and observations, and its emergent high-dimensional geometry provides a sense of direction for reaching distant goals. This quasi-Euclidean sense of direction provides a simple heuristic for online planning that works almost as well as the best offline planning algorithms from AI. If the problem space is a physical space, this method automatically extracts structural regularities from the sequence of observations that it receives so that it can generalize to unseen parts. This speeds up learning of navigation in 2D mazes and the locomotion with complex actuator systems, such as legged bodies. The cognitive map learner that we propose does not require a teacher, similar to self-attention networks (Transformers). But in contrast to Transformers, it does not require backpropagation of errors or very large datasets for learning. Hence it provides a blue-print for future energy-efficient neuromorphic hardware that acquires advanced cognitive capabilities through autonomous on-chip learning.
Collapse
Affiliation(s)
- Christoph Stöckl
- Institute of Theoretical Computer Science, Graz University of Technology, 8010, Graz, Austria
| | - Yukun Yang
- Institute of Theoretical Computer Science, Graz University of Technology, 8010, Graz, Austria
| | - Wolfgang Maass
- Institute of Theoretical Computer Science, Graz University of Technology, 8010, Graz, Austria.
| |
Collapse
|
2
|
Bybee C, Kleyko D, Nikonov DE, Khosrowshahi A, Olshausen BA, Sommer FT. Efficient optimization with higher-order ising machines. Nat Commun 2023; 14:6033. [PMID: 37758716 PMCID: PMC10533504 DOI: 10.1038/s41467-023-41214-9] [Citation(s) in RCA: 1] [Impact Index Per Article: 1.0] [Reference Citation Analysis] [Abstract] [Key Words] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 12/06/2022] [Accepted: 08/28/2023] [Indexed: 09/29/2023] Open
Abstract
A prominent approach to solving combinatorial optimization problems on parallel hardware is Ising machines, i.e., hardware implementations of networks of interacting binary spin variables. Most Ising machines leverage second-order interactions although important classes of optimization problems, such as satisfiability problems, map more seamlessly to Ising networks with higher-order interactions. Here, we demonstrate that higher-order Ising machines can solve satisfiability problems more resource-efficiently in terms of the number of spin variables and their connections when compared to traditional second-order Ising machines. Further, our results show on a benchmark dataset of Boolean k-satisfiability problems that higher-order Ising machines implemented with coupled oscillators rapidly find solutions that are better than second-order Ising machines, thus, improving the current state-of-the-art for Ising machines.
Collapse
Affiliation(s)
- Connor Bybee
- Redwood Center for Theoretical Neuroscience, University of California, Berkeley, CA, USA.
| | - Denis Kleyko
- Redwood Center for Theoretical Neuroscience, University of California, Berkeley, CA, USA
- Intelligent Systems Lab, Research Institutes of Sweden, Kista, Sweden
| | | | - Amir Khosrowshahi
- Redwood Center for Theoretical Neuroscience, University of California, Berkeley, CA, USA
- Components Research, Intel, Hillsboro, OR, USA
| | - Bruno A Olshausen
- Redwood Center for Theoretical Neuroscience, University of California, Berkeley, CA, USA
| | - Friedrich T Sommer
- Redwood Center for Theoretical Neuroscience, University of California, Berkeley, CA, USA.
- Neuromorphic Computing Lab, Intel, Santa Clara, CA, USA.
| |
Collapse
|
3
|
Liang L, Hu X, Deng L, Wu Y, Li G, Ding Y, Li P, Xie Y. Exploring Adversarial Attack in Spiking Neural Networks With Spike-Compatible Gradient. IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS 2023; 34:2569-2583. [PMID: 34473634 DOI: 10.1109/tnnls.2021.3106961] [Citation(s) in RCA: 2] [Impact Index Per Article: 2.0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 05/04/2023]
Abstract
Spiking neural network (SNN) is broadly deployed in neuromorphic devices to emulate brain function. In this context, SNN security becomes important while lacking in-depth investigation. To this end, we target the adversarial attack against SNNs and identify several challenges distinct from the artificial neural network (ANN) attack: 1) current adversarial attack is mainly based on gradient information that presents in a spatiotemporal pattern in SNNs, hard to obtain with conventional backpropagation algorithms; 2) the continuous gradient of the input is incompatible with the binary spiking input during gradient accumulation, hindering the generation of spike-based adversarial examples; and 3) the input gradient can be all-zeros (i.e., vanishing) sometimes due to the zero-dominant derivative of the firing function. Recently, backpropagation through time (BPTT)-inspired learning algorithms are widely introduced into SNNs to improve the performance, which brings the possibility to attack the models accurately given spatiotemporal gradient maps. We propose two approaches to address the above challenges of gradient-input incompatibility and gradient vanishing. Specifically, we design a gradient-to-spike (G2S) converter to convert continuous gradients to ternary ones compatible with spike inputs. Then, we design a restricted spike flipper (RSF) to construct ternary gradients that can randomly flip the spike inputs with a controllable turnover rate, when meeting all-zero gradients. Putting these methods together, we build an adversarial attack methodology for SNNs. Moreover, we analyze the influence of the training loss function and the firing threshold of the penultimate layer on the attack effectiveness. Extensive experiments are conducted to validate our solution. Besides the quantitative analysis of the influence factors, we also compare SNNs and ANNs against adversarial attacks under different attack methods. This work can help reveal what happens in SNN attacks and might stimulate more research on the security of SNN models and neuromorphic devices.
Collapse
|
4
|
Korcsak-Gorzo A, Müller MG, Baumbach A, Leng L, Breitwieser OJ, van Albada SJ, Senn W, Meier K, Legenstein R, Petrovici MA. Cortical oscillations support sampling-based computations in spiking neural networks. PLoS Comput Biol 2022; 18:e1009753. [PMID: 35324886 PMCID: PMC8947809 DOI: 10.1371/journal.pcbi.1009753] [Citation(s) in RCA: 3] [Impact Index Per Article: 1.5] [Reference Citation Analysis] [Abstract] [MESH Headings] [Grants] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 04/06/2021] [Accepted: 12/14/2021] [Indexed: 11/19/2022] Open
Abstract
Being permanently confronted with an uncertain world, brains have faced evolutionary pressure to represent this uncertainty in order to respond appropriately. Often, this requires visiting multiple interpretations of the available information or multiple solutions to an encountered problem. This gives rise to the so-called mixing problem: since all of these "valid" states represent powerful attractors, but between themselves can be very dissimilar, switching between such states can be difficult. We propose that cortical oscillations can be effectively used to overcome this challenge. By acting as an effective temperature, background spiking activity modulates exploration. Rhythmic changes induced by cortical oscillations can then be interpreted as a form of simulated tempering. We provide a rigorous mathematical discussion of this link and study some of its phenomenological implications in computer simulations. This identifies a new computational role of cortical oscillations and connects them to various phenomena in the brain, such as sampling-based probabilistic inference, memory replay, multisensory cue combination, and place cell flickering.
Collapse
Affiliation(s)
- Agnes Korcsak-Gorzo
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
- Institute of Neuroscience and Medicine (INM-6) and Institute for Advanced Simulation (IAS-6) and JARA-Institute Brain Structure-Function Relationships (INM-10), Jülich Research Centre, Jülich, Germany
- RWTH Aachen University, Aachen, Germany
| | - Michael G. Müller
- Institute of Theoretical Computer Science, Graz University of Technology, Graz, Austria
| | - Andreas Baumbach
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
- Department of Physiology, University of Bern, Bern, Switzerland
| | - Luziwei Leng
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
| | | | - Sacha J. van Albada
- Institute of Neuroscience and Medicine (INM-6) and Institute for Advanced Simulation (IAS-6) and JARA-Institute Brain Structure-Function Relationships (INM-10), Jülich Research Centre, Jülich, Germany
- Institute of Zoology, University of Cologne, Cologne, Germany
| | - Walter Senn
- Department of Physiology, University of Bern, Bern, Switzerland
| | - Karlheinz Meier
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
| | - Robert Legenstein
- Institute of Theoretical Computer Science, Graz University of Technology, Graz, Austria
| | - Mihai A. Petrovici
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
- Department of Physiology, University of Bern, Bern, Switzerland
| |
Collapse
|
5
|
Acharya SK, Galli E, Mallinson JB, Bose SK, Wagner F, Heywood ZE, Bones PJ, Arnold MD, Brown SA. Stochastic Spiking Behavior in Neuromorphic Networks Enables True Random Number Generation. ACS APPLIED MATERIALS & INTERFACES 2021; 13:52861-52870. [PMID: 34719914 DOI: 10.1021/acsami.1c13668] [Citation(s) in RCA: 6] [Impact Index Per Article: 2.0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 06/13/2023]
Abstract
There is currently a great deal of interest in the use of nanoscale devices to emulate the behaviors of neurons and synapses and to facilitate brain-inspired computation. Here, it is shown that percolating networks of nanoparticles exhibit stochastic spiking behavior that is strikingly similar to that observed in biological neurons. The spiking rate can be controlled by the input stimulus, similar to "rate coding" in biology, and the distributions of times between events are log-normal, providing insights into the atomic-scale spiking mechanism. The stochasticity of the spiking behavior is then used for true random number generation, and the high quality of the generated random bit-streams is demonstrated, opening up promising routes toward integration of neuromorphic computing with secure information processing.
Collapse
Affiliation(s)
- Susant K Acharya
- The MacDiarmid Institute for Advanced Materials and Nanotechnology, School of Physical and Chemical Sciences, Te Kura Matu, University of Canterbury, Private Bag 4800, Christchurch 8140, New Zealand
| | - Edoardo Galli
- The MacDiarmid Institute for Advanced Materials and Nanotechnology, School of Physical and Chemical Sciences, Te Kura Matu, University of Canterbury, Private Bag 4800, Christchurch 8140, New Zealand
| | - Joshua B Mallinson
- The MacDiarmid Institute for Advanced Materials and Nanotechnology, School of Physical and Chemical Sciences, Te Kura Matu, University of Canterbury, Private Bag 4800, Christchurch 8140, New Zealand
| | - Saurabh K Bose
- The MacDiarmid Institute for Advanced Materials and Nanotechnology, School of Physical and Chemical Sciences, Te Kura Matu, University of Canterbury, Private Bag 4800, Christchurch 8140, New Zealand
| | - Ford Wagner
- The MacDiarmid Institute for Advanced Materials and Nanotechnology, School of Physical and Chemical Sciences, Te Kura Matu, University of Canterbury, Private Bag 4800, Christchurch 8140, New Zealand
| | - Zachary E Heywood
- Electrical and Computer Engineering, University of Canterbury, Private Bag 4800, Christchurch 8140, New Zealand
| | - Philip J Bones
- Electrical and Computer Engineering, University of Canterbury, Private Bag 4800, Christchurch 8140, New Zealand
| | - Matthew D Arnold
- School of Mathematical and Physical Sciences, University of Technology Sydney, P.O. Box 123, Broadway, Sydney, New South Wales 2007, Australia
| | - Simon A Brown
- The MacDiarmid Institute for Advanced Materials and Nanotechnology, School of Physical and Chemical Sciences, Te Kura Matu, University of Canterbury, Private Bag 4800, Christchurch 8140, New Zealand
| |
Collapse
|
6
|
Hasani R, Ferrari G, Yamamoto H, Tanii T, Prati E. Role of Noise in Spontaneous Activity of Networks of Neurons on Patterned Silicon Emulated by Noise–activated CMOS Neural Nanoelectronic Circuits. NANO EXPRESS 2021. [DOI: 10.1088/2632-959x/abf2ae] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 11/11/2022]
Abstract
Abstract
Background noise in biological cortical microcircuits constitutes a powerful resource to assess their computational tasks, including, for instance, the synchronization of spiking activity, the enhancement of the speed of information transmission, and the minimization of the corruption of signals. We explore the correlation of spontaneous firing activity of ≈ 100 biological neurons adhering to engineered scaffolds by governing the number of functionalized patterned connection pathways among groups of neurons. We then emulate the biological system by a series of noise-activated silicon neural network simulations. We show that by suitably tuning both the amplitude of noise and the number of synapses between the silicon neurons, the same controlled correlation of the biological population is achieved. Our results extend to a realistic silicon nanoelectronics neuron design using noise injection to be exploited in artificial spiking neural networks such as liquid state machines and recurrent neural networks for stochastic computation.
Collapse
|
7
|
Spike-induced ordering: Stochastic neural spikes provide immediate adaptability to the sensorimotor system. Proc Natl Acad Sci U S A 2020; 117:12486-12496. [PMID: 32430332 PMCID: PMC7275765 DOI: 10.1073/pnas.1819707117] [Citation(s) in RCA: 6] [Impact Index Per Article: 1.5] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/18/2022] Open
Abstract
The functional advantages of using a stochastically spiking neural network (sSNN) instead of a nonspiking neural network (NS-NN) have remained largely unknown. We developed an architecture which enabled the parametric adjustment of the spikiness (i.e., impulsive dynamics and stochasticity) of the sSNN output and observed that stochastic spikes instantaneously induced the ordered motion of a dynamical system. We demonstrated the benefits of sSNNs using a musculoskeletal bipedal walker and, moreover, showed that the decrease in the spikiness of motor neuron output leads to a reduction in adaptability. Stochastic spikes may aid the adaptation of a biological system to sudden perturbations or environmental changes. Our architecture can easily be connected to the conventional NS-NN and may superimpose the on-site adaptability. Most biological neurons exhibit stochastic and spiking action potentials. However, the benefits of stochastic spikes versus continuous signals other than noise tolerance and energy efficiency remain largely unknown. In this study, we provide an insight into the potential roles of stochastic spikes, which may be beneficial for producing on-site adaptability in biological sensorimotor agents. We developed a platform that enables parametric modulation of the stochastic and discontinuous output of a stochastically spiking neural network (sSNN) to the rate-coded smooth output. This platform was applied to a complex musculoskeletal–neural system of a bipedal walker, and we demonstrated how stochastic spikes may help improve on-site adaptability of a bipedal walker to slippery surfaces or perturbation of random external forces. We further applied our sSNN platform to more general and simple sensorimotor agents and demonstrated four basic functions provided by an sSNN: 1) synchronization to a natural frequency, 2) amplification of the resonant motion in a natural frequency, 3) basin enlargement of the behavioral goal state, and 4) rapid complexity reduction and regular motion pattern formation. We propose that the benefits of sSNNs are not limited to musculoskeletal dynamics. Indeed, a wide range of the stability and adaptability of biological systems may arise from stochastic spiking dynamics.
Collapse
|
8
|
Gangopadhyay A, Mehta D, Chakrabartty S. A Spiking Neuron and Population Model Based on the Growth Transform Dynamical System. Front Neurosci 2020; 14:425. [PMID: 32477051 PMCID: PMC7235464 DOI: 10.3389/fnins.2020.00425] [Citation(s) in RCA: 3] [Impact Index Per Article: 0.8] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 02/17/2020] [Accepted: 04/07/2020] [Indexed: 11/13/2022] Open
Abstract
In neuromorphic engineering, neural populations are generally modeled in a bottom-up manner, where individual neuron models are connected through synapses to form large-scale spiking networks. Alternatively, a top-down approach treats the process of spike generation and neural representation of excitation in the context of minimizing some measure of network energy. However, these approaches usually define the energy functional in terms of some statistical measure of spiking activity (ex. firing rates), which does not allow independent control and optimization of neurodynamical parameters. In this paper, we introduce a new spiking neuron and population model where the dynamical and spiking responses of neurons can be derived directly from a network objective or energy functional of continuous-valued neural variables like the membrane potential. The key advantage of the model is that it allows for independent control over three neuro-dynamical properties: (a) control over the steady-state population dynamics that encodes the minimum of an exact network energy functional; (b) control over the shape of the action potentials generated by individual neurons in the network without affecting the network minimum; and (c) control over spiking statistics and transient population dynamics without affecting the network minimum or the shape of action potentials. At the core of the proposed model are different variants of Growth Transform dynamical systems that produce stable and interpretable population dynamics, irrespective of the network size and the type of neuronal connectivity (inhibitory or excitatory). In this paper, we present several examples where the proposed model has been configured to produce different types of single-neuron dynamics as well as population dynamics. In one such example, the network is shown to adapt such that it encodes the steady-state solution using a reduced number of spikes upon convergence to the optimal solution. In this paper, we use this network to construct a spiking associative memory that uses fewer spikes compared to conventional architectures, while maintaining high recall accuracy at high memory loads.
Collapse
Affiliation(s)
- Ahana Gangopadhyay
- Department of Electrical and Systems Engineering, Washington University in St. Louis, St. Louis, MO, United States
| | - Darshit Mehta
- Department of Biomedical Engineering, Washington University in St. Louis, St. Louis, MO, United States
| | - Shantanu Chakrabartty
- Department of Electrical and Systems Engineering, Washington University in St. Louis, St. Louis, MO, United States
| |
Collapse
|
9
|
Kungl AF, Schmitt S, Klähn J, Müller P, Baumbach A, Dold D, Kugele A, Müller E, Koke C, Kleider M, Mauch C, Breitwieser O, Leng L, Gürtler N, Güttler M, Husmann D, Husmann K, Hartel A, Karasenko V, Grübl A, Schemmel J, Meier K, Petrovici MA. Accelerated Physical Emulation of Bayesian Inference in Spiking Neural Networks. Front Neurosci 2019; 13:1201. [PMID: 31798400 PMCID: PMC6868054 DOI: 10.3389/fnins.2019.01201] [Citation(s) in RCA: 17] [Impact Index Per Article: 3.4] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 07/10/2019] [Accepted: 10/23/2019] [Indexed: 11/13/2022] Open
Abstract
The massively parallel nature of biological information processing plays an important role due to its superiority in comparison to human-engineered computing devices. In particular, it may hold the key to overcoming the von Neumann bottleneck that limits contemporary computer architectures. Physical-model neuromorphic devices seek to replicate not only this inherent parallelism, but also aspects of its microscopic dynamics in analog circuits emulating neurons and synapses. However, these machines require network models that are not only adept at solving particular tasks, but that can also cope with the inherent imperfections of analog substrates. We present a spiking network model that performs Bayesian inference through sampling on the BrainScaleS neuromorphic platform, where we use it for generative and discriminative computations on visual data. By illustrating its functionality on this platform, we implicitly demonstrate its robustness to various substrate-specific distortive effects, as well as its accelerated capability for computation. These results showcase the advantages of brain-inspired physical computation and provide important building blocks for large-scale neuromorphic applications.
Collapse
Affiliation(s)
- Akos F Kungl
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
| | - Sebastian Schmitt
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
| | - Johann Klähn
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
| | - Paul Müller
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
| | - Andreas Baumbach
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
| | - Dominik Dold
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
| | - Alexander Kugele
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
| | - Eric Müller
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
| | - Christoph Koke
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
| | - Mitja Kleider
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
| | - Christian Mauch
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
| | - Oliver Breitwieser
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
| | - Luziwei Leng
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
| | - Nico Gürtler
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
| | - Maurice Güttler
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
| | - Dan Husmann
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
| | - Kai Husmann
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
| | - Andreas Hartel
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
| | - Vitali Karasenko
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
| | - Andreas Grübl
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
| | - Johannes Schemmel
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
| | - Karlheinz Meier
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany
| | - Mihai A Petrovici
- Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany.,Department of Physiology, University of Bern, Bern, Switzerland
| |
Collapse
|
10
|
Abstract
Neuronal network models of high-level brain functions such as memory recall and reasoning often rely on the presence of some form of noise. The majority of these models assumes that each neuron in the functional network is equipped with its own private source of randomness, often in the form of uncorrelated external noise. In vivo, synaptic background input has been suggested to serve as the main source of noise in biological neuronal networks. However, the finiteness of the number of such noise sources constitutes a challenge to this idea. Here, we show that shared-noise correlations resulting from a finite number of independent noise sources can substantially impair the performance of stochastic network models. We demonstrate that this problem is naturally overcome by replacing the ensemble of independent noise sources by a deterministic recurrent neuronal network. By virtue of inhibitory feedback, such networks can generate small residual spatial correlations in their activity which, counter to intuition, suppress the detrimental effect of shared input. We exploit this mechanism to show that a single recurrent network of a few hundred neurons can serve as a natural noise source for a large ensemble of functional networks performing probabilistic computations, each comprising thousands of units.
Collapse
|
11
|
Tang J, Yuan F, Shen X, Wang Z, Rao M, He Y, Sun Y, Li X, Zhang W, Li Y, Gao B, Qian H, Bi G, Song S, Yang JJ, Wu H. Bridging Biological and Artificial Neural Networks with Emerging Neuromorphic Devices: Fundamentals, Progress, and Challenges. ADVANCED MATERIALS (DEERFIELD BEACH, FLA.) 2019; 31:e1902761. [PMID: 31550405 DOI: 10.1002/adma.201902761] [Citation(s) in RCA: 160] [Impact Index Per Article: 32.0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Subscribe] [Scholar Register] [Received: 04/30/2019] [Revised: 08/16/2019] [Indexed: 05/08/2023]
Abstract
As the research on artificial intelligence booms, there is broad interest in brain-inspired computing using novel neuromorphic devices. The potential of various emerging materials and devices for neuromorphic computing has attracted extensive research efforts, leading to a large number of publications. Going forward, in order to better emulate the brain's functions, its relevant fundamentals, working mechanisms, and resultant behaviors need to be re-visited, better understood, and connected to electronics. A systematic overview of biological and artificial neural systems is given, along with their related critical mechanisms. Recent progress in neuromorphic devices is reviewed and, more importantly, the existing challenges are highlighted to hopefully shed light on future research directions.
Collapse
Affiliation(s)
- Jianshi Tang
- Institute of Microelectronics, Beijing Innovation Center for Future Chips (ICFC), Tsinghua University, Beijing, 100084, China
- Beijing National Research Center for Information Science and Technology (BNRist), Tsinghua University, Beijing, 100084, China
| | - Fang Yuan
- Institute of Microelectronics, Beijing Innovation Center for Future Chips (ICFC), Tsinghua University, Beijing, 100084, China
| | - Xinke Shen
- Tsinghua Laboratory of Brain and Intelligence and Department of Biomedical Engineering, School of Medicine, Tsinghua University, Beijing, 100084, China
| | - Zhongrui Wang
- Department of Electrical and Computer Engineering, University of Massachusetts, Amherst, MA, 01003, USA
| | - Mingyi Rao
- Department of Electrical and Computer Engineering, University of Massachusetts, Amherst, MA, 01003, USA
| | - Yuanyuan He
- Tsinghua Laboratory of Brain and Intelligence and Department of Biomedical Engineering, School of Medicine, Tsinghua University, Beijing, 100084, China
| | - Yuhao Sun
- Tsinghua Laboratory of Brain and Intelligence and Department of Biomedical Engineering, School of Medicine, Tsinghua University, Beijing, 100084, China
| | - Xinyi Li
- Beijing National Research Center for Information Science and Technology (BNRist), Tsinghua University, Beijing, 100084, China
| | - Wenbin Zhang
- Institute of Microelectronics, Beijing Innovation Center for Future Chips (ICFC), Tsinghua University, Beijing, 100084, China
| | - Yijun Li
- Institute of Microelectronics, Beijing Innovation Center for Future Chips (ICFC), Tsinghua University, Beijing, 100084, China
| | - Bin Gao
- Institute of Microelectronics, Beijing Innovation Center for Future Chips (ICFC), Tsinghua University, Beijing, 100084, China
- Beijing National Research Center for Information Science and Technology (BNRist), Tsinghua University, Beijing, 100084, China
| | - He Qian
- Institute of Microelectronics, Beijing Innovation Center for Future Chips (ICFC), Tsinghua University, Beijing, 100084, China
- Beijing National Research Center for Information Science and Technology (BNRist), Tsinghua University, Beijing, 100084, China
| | - Guoqiang Bi
- School of Life Sciences, University of Science and Technology of China, Hefei, 230027, China
| | - Sen Song
- Tsinghua Laboratory of Brain and Intelligence and Department of Biomedical Engineering, School of Medicine, Tsinghua University, Beijing, 100084, China
| | - J Joshua Yang
- Department of Electrical and Computer Engineering, University of Massachusetts, Amherst, MA, 01003, USA
| | - Huaqiang Wu
- Institute of Microelectronics, Beijing Innovation Center for Future Chips (ICFC), Tsinghua University, Beijing, 100084, China
- Beijing National Research Center for Information Science and Technology (BNRist), Tsinghua University, Beijing, 100084, China
| |
Collapse
|
12
|
Deng L, Wu Y, Hu X, Liang L, Ding Y, Li G, Zhao G, Li P, Xie Y. Rethinking the performance comparison between SNNS and ANNS. Neural Netw 2019; 121:294-307. [PMID: 31586857 DOI: 10.1016/j.neunet.2019.09.005] [Citation(s) in RCA: 50] [Impact Index Per Article: 10.0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/02/2019] [Revised: 09/04/2019] [Accepted: 09/05/2019] [Indexed: 01/21/2023]
Abstract
Artificial neural networks (ANNs), a popular path towards artificial intelligence, have experienced remarkable success via mature models, various benchmarks, open-source datasets, and powerful computing platforms. Spiking neural networks (SNNs), a category of promising models to mimic the neuronal dynamics of the brain, have gained much attention for brain inspired computing and been widely deployed on neuromorphic devices. However, for a long time, there are ongoing debates and skepticisms about the value of SNNs in practical applications. Except for the low power attribute benefit from the spike-driven processing, SNNs usually perform worse than ANNs especially in terms of the application accuracy. Recently, researchers attempt to address this issue by borrowing learning methodologies from ANNs, such as backpropagation, to train high-accuracy SNN models. The rapid progress in this domain continuously produces amazing results with ever-increasing network size, whose growing path seems similar to the development of deep learning. Although these ways endow SNNs the capability to approach the accuracy of ANNs, the natural superiorities of SNNs and the way to outperform ANNs are potentially lost due to the use of ANN-oriented workloads and simplistic evaluation metrics. In this paper, we take the visual recognition task as a case study to answer the questions of "what workloads are ideal for SNNs and how to evaluate SNNs makes sense". We design a series of contrast tests using different types of datasets (ANN-oriented and SNN-oriented), diverse processing models, signal conversion methods, and learning algorithms. We propose comprehensive metrics on the application accuracy and the cost of memory & compute to evaluate these models, and conduct extensive experiments. We evidence the fact that on ANN-oriented workloads, SNNs fail to beat their ANN counterparts; while on SNN-oriented workloads, SNNs can fully perform better. We further demonstrate that in SNNs there exists a trade-off between the application accuracy and the execution cost, which will be affected by the simulation time window and firing threshold. Based on these abundant analyses, we recommend the most suitable model for each scenario. To the best of our knowledge, this is the first work using systematical comparisons to explicitly reveal that the straightforward workload porting from ANNs to SNNs is unwise although many works are doing so and a comprehensive evaluation indeed matters. Finally, we highlight the urgent need to build a benchmarking framework for SNNs with broader tasks, datasets, and metrics.
Collapse
Affiliation(s)
- Lei Deng
- Department of Precision Instrument, Center for Brain Inspired Computing Research, Tsinghua University, Beijing 100084, China; Department of Electrical and Computer Engineering, University of California, Santa Barbara,, CA 93106, USA.
| | - Yujie Wu
- Department of Precision Instrument, Center for Brain Inspired Computing Research, Tsinghua University, Beijing 100084, China.
| | - Xing Hu
- Department of Electrical and Computer Engineering, University of California, Santa Barbara,, CA 93106, USA.
| | - Ling Liang
- Department of Electrical and Computer Engineering, University of California, Santa Barbara,, CA 93106, USA.
| | - Yufei Ding
- Department of Computer Science, University of California, Santa Barbara,, CA 93106, USA.
| | - Guoqi Li
- Department of Precision Instrument, Center for Brain Inspired Computing Research, Tsinghua University, Beijing 100084, China.
| | - Guangshe Zhao
- School of Electronic and Information Engineering, Xi'an Jiaotong University, Xi'an 710049, China.
| | - Peng Li
- Department of Electrical and Computer Engineering, University of California, Santa Barbara,, CA 93106, USA.
| | - Yuan Xie
- Department of Electrical and Computer Engineering, University of California, Santa Barbara,, CA 93106, USA.
| |
Collapse
|
13
|
Nobukawa S, Nishimura H, Yamanishi T. Temporal-specific complexity of spiking patterns in spontaneous activity induced by a dual complex network structure. Sci Rep 2019; 9:12749. [PMID: 31484990 PMCID: PMC6726653 DOI: 10.1038/s41598-019-49286-8] [Citation(s) in RCA: 14] [Impact Index Per Article: 2.8] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 11/14/2018] [Accepted: 08/22/2019] [Indexed: 11/08/2022] Open
Abstract
Temporal fluctuation of neural activity in the brain has an important function in optimal information processing. Spontaneous activity is a source of such fluctuation. The distribution of excitatory postsynaptic potentials (EPSPs) between cortical pyramidal neurons can follow a log-normal distribution. Recent studies have shown that networks connected by weak synapses exhibit characteristics of a random network, whereas networks connected by strong synapses have small-world characteristics of small path lengths and large cluster coefficients. To investigate the relationship between temporal complexity spontaneous activity and structural network duality in synaptic connections, we executed a simulation study using the leaky integrate-and-fire spiking neural network with log-normal synaptic weight distribution for the EPSPs and duality of synaptic connectivity, depending on synaptic weight. We conducted multiscale entropy analysis of the temporal spiking activity. Our simulation demonstrated that, when strong synaptic connections approach a small-world network, specific spiking patterns arise during irregular spatio-temporal spiking activity, and the complexity at the large temporal scale (i.e., slow frequency) is enhanced. Moreover, we confirmed through a surrogate data analysis that slow temporal dynamics reflect a deterministic process in the spiking neural networks. This modelling approach may improve the understanding of the spatio-temporal complex neural activity in the brain.
Collapse
Affiliation(s)
- Sou Nobukawa
- Department of Computer Science, Chiba Institute of Technology, 2-17-1 Tsudanuma, Narashino, Chiba, 275-0016, Japan.
| | - Haruhiko Nishimura
- Graduate School of Applied Informatics, University of Hyogo, 7-1-28 Chuo-ku, Kobe, Hyogo, 650-8588, Japan
| | - Teruya Yamanishi
- AI & IoT Center, Department of Management and Information Sciences, Fukui University of Technology, 3-6-1 Gakuen, Fukui, 910-8505, Japan
| |
Collapse
|
14
|
Fang Y, Wang Z, Gomez J, Datta S, Khan AI, Raychowdhury A. A Swarm Optimization Solver Based on Ferroelectric Spiking Neural Networks. Front Neurosci 2019; 13:855. [PMID: 31456659 PMCID: PMC6700359 DOI: 10.3389/fnins.2019.00855] [Citation(s) in RCA: 11] [Impact Index Per Article: 2.2] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 03/07/2019] [Accepted: 07/30/2019] [Indexed: 11/13/2022] Open
Abstract
As computational models inspired by the biological neural system, spiking neural networks (SNN) continue to demonstrate great potential in the landscape of artificial intelligence, particularly in tasks such as recognition, inference, and learning. While SNN focuses on achieving high-level intelligence of individual creatures, Swarm Intelligence (SI) is another type of bio-inspired models that mimic the collective intelligence of biological swarms, i.e., bird flocks, fish school and ant colonies. SI algorithms provide efficient and practical solutions to many difficult optimization problems through multi-agent metaheuristic search. Bridging these two distinct subfields of artificial intelligence has the potential to harness collective behavior and learning ability of biological systems. In this work, we explore the feasibility of connecting these two models by implementing a generalized SI model on SNN. In the proposed computing paradigm, we use SNNs to represent agents in the swarm and encode problem solutions with the spike firing rate and with spike timing. The coupled neurons communicate and modulate each other's action potentials through event-driven spikes and synchronize their dynamics around the states of optimal solutions. We demonstrate that such an SI-SNN model is capable of efficiently solving optimization problems, such as parameter optimization of continuous functions and a ubiquitous combinatorial optimization problem, namely, the traveling salesman problem with near-optimal solutions. Furthermore, we demonstrate an efficient implementation of such neural dynamics on an emerging hardware platform, namely ferroelectric field-effect transistor (FeFET) based spiking neurons. Such an emerging in-silico neuron is composed of a compact 1T-1FeFET structure with both excitatory and inhibitory inputs. We show that the designed neuromorphic system can serve as an optimization solver with high-performance and high energy-efficiency.
Collapse
Affiliation(s)
- Yan Fang
- School of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA, United States
| | - Zheng Wang
- School of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA, United States
| | - Jorge Gomez
- Department of Electrical Engineering, University of Notre Dame, Notre Dame, IN, United States
| | - Suman Datta
- Department of Electrical Engineering, University of Notre Dame, Notre Dame, IN, United States
| | - Asif I Khan
- School of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA, United States
| | - Arijit Raychowdhury
- School of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA, United States
| |
Collapse
|
15
|
Krichmar JL, Severa W, Khan MS, Olds JL. Making BREAD: Biomimetic Strategies for Artificial Intelligence Now and in the Future. Front Neurosci 2019; 13:666. [PMID: 31316340 PMCID: PMC6610536 DOI: 10.3389/fnins.2019.00666] [Citation(s) in RCA: 11] [Impact Index Per Article: 2.2] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 04/01/2019] [Accepted: 06/11/2019] [Indexed: 11/24/2022] Open
Abstract
The Artificial Intelligence (AI) revolution foretold of during the 1960s is well underway in the second decade of the twenty first century. Its period of phenomenal growth likely lies ahead. AI-operated machines and technologies will extend the reach of Homo sapiens far beyond the biological constraints imposed by evolution: outwards further into deep space, as well as inwards into the nano-world of DNA sequences and relevant medical applications. And yet, we believe, there are crucial lessons that biology can offer that will enable a prosperous future for AI. For machines in general, and for AI's especially, operating over extended periods or in extreme environments will require energy usage orders of magnitudes more efficient than exists today. In many operational environments, energy sources will be constrained. The AI's design and function may be dependent upon the type of energy source, as well as its availability and accessibility. Any plans for AI devices operating in a challenging environment must begin with the question of how they are powered, where fuel is located, how energy is stored and made available to the machine, and how long the machine can operate on specific energy units. While one of the key advantages of AI use is to reduce the dimensionality of a complex problem, the fact remains that some energy is required for functionality. Hence, the materials and technologies that provide the needed energy represent a critical challenge toward future use scenarios of AI and should be integrated into their design. Here we look to the brain and other aspects of biology as inspiration for Biomimetic Research for Energy-efficient AI Designs (BREAD).
Collapse
Affiliation(s)
- Jeffrey L. Krichmar
- Departments of Cognitive Sciences and Computer Science, University of California, Irvine, Irvine, CA, United States
| | - William Severa
- Sandia National Laboratories, Data-Driven and Neural Computing, Albuquerque, NM, United States
| | - Muhammad S. Khan
- Schar School, George Mason University, Arlington, VA, United States
| | - James L. Olds
- Schar School, George Mason University, Arlington, VA, United States
| |
Collapse
|
16
|
Wijesinghe P, Liyanagedera C, Roy K. Analog Approach to Constraint Satisfaction Enabled by Spin Orbit Torque Magnetic Tunnel Junctions. Sci Rep 2018; 8:6940. [PMID: 29720596 PMCID: PMC5932068 DOI: 10.1038/s41598-018-24877-z] [Citation(s) in RCA: 2] [Impact Index Per Article: 0.3] [Reference Citation Analysis] [Abstract] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 11/20/2017] [Accepted: 04/09/2018] [Indexed: 11/09/2022] Open
Abstract
Boolean satisfiability (k-SAT) is an NP-complete (k ≥ 3) problem that constitute one of the hardest classes of constraint satisfaction problems. In this work, we provide a proof of concept hardware based analog k-SAT solver, that is built using Magnetic Tunnel Junctions (MTJs). The inherent physics of MTJs, enhanced by device level modifications, is harnessed here to emulate the intricate dynamics of an analog satisfiability (SAT) solver. In the presence of thermal noise, the MTJ based system can successfully solve Boolean satisfiability problems. Most importantly, our results exhibit that, the proposed MTJ based hardware SAT solver is capable of finding a solution to a significant fraction (at least 85%) of hard 3-SAT problems, within a time that has a polynomial relationship with the number of variables(<50).
Collapse
Affiliation(s)
- Parami Wijesinghe
- Purdue University, School of Electrical and Computer Engineering, West Lafayette, Indiana, 47907, USA.
| | - Chamika Liyanagedera
- Purdue University, School of Electrical and Computer Engineering, West Lafayette, Indiana, 47907, USA
| | - Kaushik Roy
- Purdue University, School of Electrical and Computer Engineering, West Lafayette, Indiana, 47907, USA
| |
Collapse
|
17
|
Rutishauser U, Slotine JJ, Douglas RJ. Solving Constraint-Satisfaction Problems with Distributed Neocortical-Like Neuronal Networks. Neural Comput 2018; 30:1359-1393. [PMID: 29566357 PMCID: PMC5930080 DOI: 10.1162/neco_a_01074] [Citation(s) in RCA: 11] [Impact Index Per Article: 1.8] [Reference Citation Analysis] [Abstract] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 02/03/2023]
Abstract
Finding actions that satisfy the constraints imposed by both external inputs and internal representations is central to decision making. We demonstrate that some important classes of constraint satisfaction problems (CSPs) can be solved by networks composed of homogeneous cooperative-competitive modules that have connectivity similar to motifs observed in the superficial layers of neocortex. The winner-take-all modules are sparsely coupled by programming neurons that embed the constraints onto the otherwise homogeneous modular computational substrate. We show rules that embed any instance of the CSP's planar four-color graph coloring, maximum independent set, and sudoku on this substrate and provide mathematical proofs that guarantee these graph coloring problems will convergence to a solution. The network is composed of nonsaturating linear threshold neurons. Their lack of right saturation allows the overall network to explore the problem space driven through the unstable dynamics generated by recurrent excitation. The direction of exploration is steered by the constraint neurons. While many problems can be solved using only linear inhibitory constraints, network performance on hard problems benefits significantly when these negative constraints are implemented by nonlinear multiplicative inhibition. Overall, our results demonstrate the importance of instability rather than stability in network computation and offer insight into the computational role of dual inhibitory mechanisms in neural circuits.
Collapse
Affiliation(s)
- Ueli Rutishauser
- Computation and Neural Systems, Division of Biology and Biological Engineering, California Institute of Technology, Pasadena, CA 91125, U.S.A., and Cedars-Sinai Medical Center, Departments of Neurosurgery, Neurology and Biomedical Sciences, Los Angeles, CA 90048, U.S.A.
| | - Jean-Jacques Slotine
- Nonlinear Systems Laboratory, Department of Mechanical Engineering and Department of Brain and Cognitive Sciences, MIT, Cambridge, MA 02139, U.S.A.
| | - Rodney J Douglas
- Institute of Neuroinformatics, University and ETH Zurich, Zurich 8057, Switzerland
| |
Collapse
|
18
|
Fonseca Guerra GA, Furber SB. Using Stochastic Spiking Neural Networks on SpiNNaker to Solve Constraint Satisfaction Problems. Front Neurosci 2017; 11:714. [PMID: 29311791 PMCID: PMC5742150 DOI: 10.3389/fnins.2017.00714] [Citation(s) in RCA: 8] [Impact Index Per Article: 1.1] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 08/28/2017] [Accepted: 12/06/2017] [Indexed: 11/13/2022] Open
Abstract
Constraint satisfaction problems (CSP) are at the core of numerous scientific and technological applications. However, CSPs belong to the NP-complete complexity class, for which the existence (or not) of efficient algorithms remains a major unsolved question in computational complexity theory. In the face of this fundamental difficulty heuristics and approximation methods are used to approach instances of NP (e.g., decision and hard optimization problems). The human brain efficiently handles CSPs both in perception and behavior using spiking neural networks (SNNs), and recent studies have demonstrated that the noise embedded within an SNN can be used as a computational resource to solve CSPs. Here, we provide a software framework for the implementation of such noisy neural solvers on the SpiNNaker massively parallel neuromorphic hardware, further demonstrating their potential to implement a stochastic search that solves instances of P and NP problems expressed as CSPs. This facilitates the exploration of new optimization strategies and the understanding of the computational abilities of SNNs. We demonstrate the basic principles of the framework by solving difficult instances of the Sudoku puzzle and of the map color problem, and explore its application to spin glasses. The solver works as a stochastic dynamical system, which is attracted by the configuration that solves the CSP. The noise allows an optimal exploration of the space of configurations, looking for the satisfiability of all the constraints; if applied discontinuously, it can also force the system to leap to a new random configuration effectively causing a restart.
Collapse
Affiliation(s)
- Gabriel A Fonseca Guerra
- Advanced Processor Technologies Group, School of Computer Science, University of Manchester, Manchester, United Kingdom
| | - Steve B Furber
- Advanced Processor Technologies Group, School of Computer Science, University of Manchester, Manchester, United Kingdom
| |
Collapse
|