Neuro-CIM: ADC-Less Neuromorphic Computing-in-Memory Processor With Operation Gating/Stopping and Digital-Analog Networks

Article Computer Science, Artificial Intelligence

A Tandem Learning Rule for Effective Training and Rapid Inference of Deep Spiking Neural Networks

Jibin Wu et al.

Summary: Spiking neural networks (SNNs) are a prominent biologically inspired computing model but are not directly applicable to standard error backpropagation algorithm due to the nondifferentiable nature of spiking neuronal functions. In this work, a tandem learning framework consisting of an SNN and an artificial neural network (ANN) is proposed to train the SNN at the spike-train level. The proposed tandem learning rule demonstrates competitive pattern recognition and regression capabilities with reduced inference time and total synaptic operations.

IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS (2023)

Add to Collection

Article Computer Science, Artificial Intelligence

Progressive Tandem Learning for Pattern Recognition With Deep Spiking Neural Networks

Jibin Wu et al.

Summary: This paper proposes a novel framework for rapid and efficient pattern recognition in spiking neural networks (SNNs), which involves an ANN-to-SNN conversion and layer-wise learning approach called progressive tandem learning. By studying the equivalence between ANNs and SNNs, a spike count-based approximation method is introduced to convert ANN neurons into SNN neurons. Additionally, a layer-wise learning method with adaptive training scheduler is used to fine-tune the network weights and compensate for approximation errors. The framework also allows hardware constraints to be progressively imposed during training and demonstrates remarkable performance on various tasks.

IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE (2022)

Add to Collection

Proceedings Paper Engineering, Electrical & Electronic