Building a Comprehensive Neuromorphic Platform for Remote Computation

Remote sensing and extreme environments present a unique and critical algorithm and hardware tradeoff due to extreme size, weight and power constraints. Consequently, in many applications, systems favor centralized computation over remote computation. However, in some real-time systems (e.g. a Mars rover) latency and other communication bottlenecks force on-board processing. With traditional processor performance at a plateau, we look to brain-inspired, non-Von Neumann neuromorphic architectures to enable future capabilities, such as event detection/tracking and intelligent decision making. These cutting-edge hardware platforms generally operate at vastly improved performance-per-Watt ratios, but have suffered from niche applications, difficult interfaces, and poor integration with existing algorithms. In this paper, we discuss methods, motivated by recent results, to produce a cohesive neuromorphic system that effectively integrates novel and traditional algorithms for context-driven remote computation. Keywords—Deep Learning; Neuromorphic Computing; Remote Computation; Neuromodulation

[1]  Ojas Parekh,et al.  Constant-Depth and Subcubic-Size Threshold Circuits for Matrix Multiplication , 2018, SPAA.

[2]  Hong Wang,et al.  Loihi: A Neuromorphic Manycore Processor with On-Chip Learning , 2018, IEEE Micro.

[3]  Yoshua Bengio,et al.  How transferable are features in deep neural networks? , 2014, NIPS.

[4]  Steve Furber,et al.  Large-scale neuromorphic computing systems , 2016, Journal of neural engineering.

[5]  Vladlen Koltun,et al.  Learning to Act by Predicting the Future , 2016, ICLR.

[6]  P. Lichtsteiner,et al.  Toward real-time particle tracking using an event-based dynamic vision sensor , 2011 .

[7]  Emre O. Neftci,et al.  Data and Power Efficient Intelligence with Neuromorphic Learning Machines , 2018, iScience.

[8]  Catherine D. Schuman,et al.  A Survey of Neuromorphic Computing and Neural Networks in Hardware , 2017, ArXiv.

[9]  Fred Rothganger,et al.  Computing with Spikes: The Advantage of Fine-Grained Timing , 2018, Neural Computation.

[10]  Tobi Delbrück,et al.  A Low Power, Fully Event-Based Gesture Recognition System , 2017, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[11]  Alex Krizhevsky,et al.  Learning Multiple Layers of Features from Tiny Images , 2009 .

[12]  Andrew S. Cassidy,et al.  Cognitive computing programming paradigm: A Corelet Language for composing networks of neurosynaptic cores , 2013, The 2013 International Joint Conference on Neural Networks (IJCNN).

[13]  Steve B. Furber,et al.  Memory-Efficient Deep Learning on a SpiNNaker 2 Prototype , 2018, Front. Neurosci..

[14]  John V. Monaco,et al.  Integer factorization with a neuromorphic sieve , 2017, 2017 IEEE International Symposium on Circuits and Systems (ISCAS).

[15]  Peter Blouw,et al.  Benchmarking Keyword Spotting Efficiency on Neuromorphic Hardware , 2018, NICE '19.

[16]  Ryad Benosman,et al.  STICK: Spike Time Interval Computational Kernel, a Framework for General Purpose Computation Using Neurons, Precise Timing, Delays, and Synchrony , 2015, Neural Computation.

[17]  W. Newsome,et al.  Context-dependent computation by recurrent dynamics in prefrontal cortex , 2013, Nature.

[18]  Andrew S. Cassidy,et al.  A million spiking-neuron integrated circuit with a scalable communication network and interface , 2014, Science.

[19]  Eugenio Culurciello,et al.  Activity-driven, event-based vision sensors , 2010, Proceedings of 2010 IEEE International Symposium on Circuits and Systems.

[20]  Bernabé Linares-Barranco,et al.  A 128$\,\times$ 128 1.5% Contrast Sensitivity 0.9% FPN 3 µs Latency 4 mW Asynchronous Frame-Free Dynamic Vision Sensor Using Transimpedance Preamplifiers , 2013, IEEE Journal of Solid-State Circuits.

[21]  Andrew S. Cassidy,et al.  Convolutional networks for fast, energy-efficient neuromorphic computing , 2016, Proceedings of the National Academy of Sciences.

[22]  Ojas Parekh,et al.  Spiking network algorithms for scientific computing , 2016, 2016 IEEE International Conference on Rebooting Computing (ICRC).

[23]  Ojas Parekh,et al.  Spiking Neural Algorithms for Markov Process Random Walk , 2018, 2018 International Joint Conference on Neural Networks (IJCNN).

[24]  James B. Aimone,et al.  Context-modulation of hippocampal dynamics and deep convolutional networks , 2017, ArXiv.

[25]  Pierre Yger,et al.  PyNN: A Common Interface for Neuronal Network Simulators , 2008, Front. Neuroinform..

[26]  Ojas Parekh,et al.  Neural computing for scientific computing applications: more than just machine learning , 2017, NCS.

[27]  Jim D. Garside,et al.  Overview of the SpiNNaker System Architecture , 2013, IEEE Transactions on Computers.

[28]  Chris Eliasmith,et al.  Training Spiking Deep Networks for Neuromorphic Hardware , 2016, ArXiv.

[29]  Craig M. Vineyard,et al.  Training deep neural networks for binary communication with the Whetstone method , 2019 .

[30]  Gert Cauwenberghs,et al.  Large-Scale Neuromorphic Spiking Array Processors: A Quest to Mimic the Brain , 2018, Front. Neurosci..

[31]  Ken Lang,et al.  NewsWeeder: Learning to Filter Netnews , 1995, ICML.