Spiking neural networks (SNNs) are a promising candidate for biologically-inspired and energy efficient computation. However, their simulation is restrictively time consuming, and creates a bottleneck in developing competitive training methods with potential deployment on neuromorphic hardware platforms, even on simple tasks. To address this issue, we provide an implementation of mini-batch processing applied to clock-based SNN simulation, leading to drastically increased data throughput. To our knowledge, this is the first general-purpose implementation of mini-batch processing in a spiking neural networks simulator, which works with arbitrary neuron and synapse models. We demonstrate nearly constant-time scaling with batch size on a simulation setup (up to GPU memory limits), and showcase the effectiveness of large batch sizes in two SNN application domains, resulting in ≈880X and ≈24X reductions in wall-clock time respectively. Different parameter reduction techniques are shown to produce different learning outcomes in a simulation of networks trained with spike-timing-dependent plasticity. Machine learning practitioners and biological modelers alike may benefit from the drastically reduced simulation time and increased iteration speed this method enables.
Paper
Full text
Minibatch Processing for Speed-up and Scalability of Spiking Neural Network Simulation
Semantic Scholar · Computer Science · 2020
Abstract
Spiking neural networks (SNNs) are a promising candidate for biologically-inspired and energy efficient computation. However, their simulation is restrictively time consuming, and creates a bottleneck in developing competitive training methods with potential deployment on neuromorphic hardware platforms, even on simple tasks. To address this issue, we provide an implementation of mini-batch processing applied to clock-based SNN simulation, leading to drastically increased data throughput. To our knowledge, this is the first general-purpose implementation of mini-batch processing in a spiking neural networks simulator, which works with arbitrary neuron and synapse models. We demonstrate nearly constant-time scaling with batch size on a simulation setup (up to GPU memory limits), and showcase the effectiveness of large batch sizes in two SNN application domains, resulting in ≈880X and ≈24X reductions in wall-clock time respectively. Different parameter reduction techniques are shown to produce different learning outcomes in a simulation of networks trained with spike-timing-dependent plasticity. Machine learning practitioners and biological modelers alike may benefit from the drastically reduced simulation time and increased iteration speed this method enables.