The field of neuromorphic computing is rapidly advancing with a focus on developing more efficient and accurate spiking neural networks (SNNs). Researchers are exploring new methods to convert artificial neural networks (ANNs) to SNNs, such as using spike accumulation and adaptive layerwise activation, which can significantly reduce the number of inference timesteps required. Additionally, neuromorphic computing is being applied to various applications, including personalized treatment for drug-resistant epilepsy, smart prosthetics, and diagnostics. Noteworthy papers in this area include PASCAL, which proposes a method for precise and efficient ANN-SNN conversion, and TS-SNN, which introduces a temporal shift module to integrate past, present, and future spike features within a single timestep. Overall, the field is moving towards more energy-efficient and biologically plausible neural networks that can be applied to a wide range of applications.
Neuromorphic Computing Advances
Sources
PASCAL: Precise and Efficient ANN- SNN Conversion using Spike Accumulation and Adaptive Layerwise Activation
Closed-loop control of seizure activity via real-time seizure forecasting by reservoir neuromorphic computing
Machine-Learning-Powered Neural Interfaces for Smart Prosthetics and Diagnostics
Ultra-Low-Power Spiking Neurons in 7 nm FinFET Technology: A Comparative Analysis of Leaky Integrate-and-Fire, Morris-Lecar, and Axon-Hillock Architectures
Izhikevich-Inspired Temporal Dynamics for Enhancing Privacy, Efficiency, and Transferability in Spiking Neural Networks
Scalable 49-Channel Neural Recorder with an Event-Driven Ramp ADC and PCA Compression in 28 nm CMOS
TS-SNN: Temporal Shift Module for Spiking Neural Networks
Threshold Modulation for Online Test-Time Adaptation of Spiking Neural Networks