Aaron R. Voelker
PhD
I started my masters in Computer Science at the lab in September 2013, and fast-tracked towards a PhD in September 2014. I successfully defended my PhD, titled "Dynamical Systems in Spiking Neuromorphic Hardware", in April 2019. I received my BMath Double Honours in Computer Science and Combinatorics & Optimization from the University of Waterloo, have worked at Amazon, Google, and Wish, and currently work for Applied Brain Research Inc.
My thesis addresses our understanding of how the brain can effectively represent and process dynamic stimuli, while using only spikes to transmit information between its billions of neurons. The Neural Engineering Framework (NEF) is a perfect fit for exploring the range of ways this can be accomplished using biologically plausible spiking neurons and synapses. We have extended the NEF to account for more detailed models of neurons and synapses, in order to deploy the NEF on neuromorphic architectures such as Braindrop and Loihi, and to better understand the class of computations made available in neural substrate. Together, this understanding is being used to build large-scale models that process dynamic stimuli online into structured representations, for use in classification, association, and prediction.
Publications
Theses
- (2019) Dynamical Systems in Spiking Neuromorphic Hardware. PhD thesis, University of Waterloo. Abstract PDF External link
Books and Book Chapters
- (2021) Programming Neuromorphics Using the Neural Engineering Framework. Nitish V. Thakor, editor. Programming Neuromorphics Using the Neural Engineering Framework, pages 1–43. Springer Singapore, Singapore. Abstract PDF DOI
Journal Articles
- (2021) Simulating and Predicting Dynamical Systems With Spatial Semantic Pointers. Neural Computation, 33(8):2033-2067. Abstract PDF DOI
- (2019) Braindrop: A Mixed-Signal Neuromorphic Architecture With a Dynamical Systems-Based Programming Model. Proceedings of the IEEE, 107:144–164. Abstract PDF DOI External link
- (2018) Improving Spiking Dynamical Networks: Accurate Delays, Higher-Order Synapses, and Time Cells. Neural Computation, 30(3):569-609. Abstract PDF DOI External link
- (2017) A neural model of hierarchical reinforcement learning. PLoS ONE, 12(7):1–39. Abstract PDF DOI
- (2016) Human-Inspired Neurorobotic System for Classifying Surface Textures by Touch. Robotics and Automation Letters, 1(1):516-523. Abstract PDF DOI
- (2016) Improving With Practice: A Neural Model of Mathematical Development. Topics in Cognitive Science. Abstract PDF DOI
- (2014) Nengo: A Python tool for building large-scale functional brain models. Frontiers in Neuroinformatics. Abstract PDF DOI External link
Conference and Workshop Papers
- (2021) Hardware Aware Training for Efficient Keyword Spotting on General Purpose and Specialized Hardware. In TinyML Research Symposium. Abstract arXiv
- (2021) A neurocomputational model of prospective and retrospective timing. In International Conference on Cognitive Modeling. Toronto, ON. Society for Mathematical Psychology. Abstract External link
- (2019) Flexible Timing with Delay Networks – The Scalar Property and Neural Scaling. In 17th Annual Meeting of the International Conference on Cognitive Modelling (ICCM). Abstract PDF External link
- (2019) A Neurocomputational Account of Ecologically Plausible, Flexible Timing with Legendre Memory. In 2nd Annual Conference of the Timing Research Forum. Querétaro, México. Abstract Poster External link
- (2019) Legendre Memory Units: Continuous-Time Representation in Recurrent Neural Networks. In Advances in Neural Information Processing Systems, 15544–15553. Abstract PDF Poster External link
- (2019) A neural representation of continuous space using fractional binding. In 41st Annual Meeting of the Cognitive Science Society. Montreal, QC. Cognitive Science Society. Abstract PDF
- (2019) Representing spatial relations with fractional binding. In 41st Annual Meeting of the Cognitive Science Society. Montreal, QC. Cognitive Science Society. Abstract PDF
- (2018) Nonlinear synaptic interaction as a computational resource in the Neural Engineering Framework. In Cosyne Abstracts. Denver USA. Abstract PDF Poster External link
- (2017) A Population-Level Approach to Temperature Robustness in Neuromorphic Systems. In IEEE International Symposium on Circuits and Systems (ISCAS). Baltimore, MD. IEEE. Abstract PDF External link
- (2017) A Spiking Independent Accumulator Model for Winner-Take-All Computation. In Proceedings of the 39th Annual Conference of the Cognitive Science Society. London, UK. Cognitive Science Society. Abstract PDF Poster External link
- (2017) A Spiking Neural Bayesian Model of Life Span Inference. In Proceedings of the 39th Annual Conference of the Cognitive Science Society. London, UK. Cognitive Science Society. Abstract PDF External link
- (2017) Extending the Neural Engineering Framework for Nonideal Silicon Synapses. In IEEE International Symposium on Circuits and Systems (ISCAS). Baltimore, MD. IEEE. Abstract PDF Poster External link
- (2016) Improving with Practice: A Neural Model of Mathematical Development. In Anna Papafragou Dan Grodner, Dan Mirman and John Trueswell, editors, Proceedings of the 38th Annual Conference of the Cognitive Science Society, 2012–2026. Philadelphia, Pennsylvania. Cognitive Science Society. Abstract PDF External link
- (2016) Efficient SpiNNaker simulation of a heteroassociative memory using the Neural Engineering Framework. In The 2016 International Joint Conference on Neural Networks (IJCNN). Vancouver, British Columbia. IEEE. Abstract External link
- (2015) Computing with temporal representations using recurrently connected populations of spiking neurons. In Connecting Network Architecture and Network Computation. Banff International Research Station for Mathematical Innovation and Discovery. Abstract DOI
- (2014) Learning large-scale heteroassociative memories in spiking neurons. In Steffen Kopecki Oscar H. Ibarra, Lila Kari, editor, Unconventional Computation and Natural Computation. London, Ontario. Springer International Publishing. Abstract PDF Poster
Technical Reports and Preprints
- (2021) Language Modeling using LMUs: 10x Better Data Efficiency or Improved Scaling Compared to Transformers. arXiv preprint. Abstract PDF arXiv
- (2020) A Spike in Performance: Training Hybrid-Spiking Neural Networks with Quantized Activation Functions. arxiv. Abstract PDF DOI
- (2017) Efficiently sampling vectors and coordinates from the n-sphere and n-ball. Technical Report, Centre for Theoretical Neuroscience, Waterloo, ON. Abstract PDF DOI External link
- (2017) Methods for applying the Neural Engineering Framework to neuromorphic hardware. arXiv preprint arXiv:1708.08133. Abstract PDF arXiv
- (2017) Analysis of oscillatory weight changes from online learning with filtered spiking feedback. Technical Report, Centre for Theoretical Neuroscience, Waterloo, ON. Abstract DOI External link
- (2017) Point Neurons with Conductance-Based Synapses in the Neural Engineering Framework. arXiv preprint arXiv:1710.07659. Abstract PDF arXiv
- (2015) A Solution to the Dynamics of the Prescribed Error Sensitivity Learning Rule. Technical Report, Centre for Theoretical Neuroscience, Waterloo, ON. Abstract PDF DOI External link