Neuromorphic Computing
Overview
Neuromorphic computing studies algorithms and physical systems that compute through neuron-like dynamics, spikes, sparse events, and sometimes sensing integrated with processing. Common approaches include spiking neural networks trained with surrogate gradients or converted from conventional networks, local learning such as spike-timing-dependent plasticity, reservoir computing, and specialized CMOS, FPGA, memristor, photonic, or analog hardware. These systems are tested on event-camera vision, speech, time-series, robotics, and edge sensing, with benchmarks such as DVS Gesture, N-MNIST, SHD, and CIFAR variants. The central challenge is turning sparse activity and device-level efficiency into reliable end-to-end gains without sacrificing accuracy, scalability, or fair comparison with conventional hardware.
Latest papers
206 papers17 papers in the last four weeks, level with the four weeks before. 0.1% of all new papers.
A Voltage-controlled MTJ-CMOS Neuron Emulating Tunable Izhikevich-Inspired Dynamics
Can Spiking Neural Networks play pinball? A neuromorphic motion detector for target tracking
NAVIR: Neuromorphic Audio-Visual Speech Recognition for Robust Human-Robot Interaction on Edge Hardware
Sensing to Intelligence: Principles for Neuromorphic Circuits and Systems
DVA-Neurons: Design and Verification of Adaptive LIF Neurons: From Single-Neuron Dynamics to Multi-Neuron Spiking Networks
A Spatiotemporal Extension of the Neuromorphic DBSCAN Implementation
flat'' and systolic''. The flat'' construction prioritizes throughput, while the systolic'' construction trades time for space resulting in a smaller, more hardware-friendly architecture at the cost of throughput. In this work, we offer spatiotemporal extensions of these two constructions to better leverage the spatiotemporal nature of event sensor data. Moreover, as in our prior work, we discuss partial or segmented implementations that further leverage time for space when hardware resources are constrained. All network constructions are provided as open-source implementations.