# Binary Weight Multi-Bit Activation Quantization for Compute-in-Memory CNN Accelerators

## TL;DR

This paper proposes a novel binary weight multi-bit activation quantization method for CNNs on compute-in-memory accelerators, improving accuracy while maintaining hardware efficiency.

## Contribution

It introduces closed-form weight quantization solutions and a differentiable activation quantization function, enhancing binarized weights and multi-bit activations for CIM platforms.

## Key findings

- Achieves 1.44%-5.46% accuracy gain on CIFAR-10 and ImageNet
- 4-bit activation quantization offers optimal hardware-performance balance
- Significantly improves binarized weight representational capacity

## Abstract

Compute-in-memory (CIM) accelerators have emerged as a promising way for enhancing the energy efficiency of convolutional neural networks (CNNs). Deploying CNNs on CIM platforms generally requires quantization of network weights and activations to meet hardware constraints. However, existing approaches either prioritize hardware efficiency with binary weight and activation quantization at the cost of accuracy, or utilize multi-bit weights and activations for greater accuracy but limited efficiency. In this paper, we introduce a novel binary weight multi-bit activation (BWMA) method for CNNs on CIM-based accelerators. Our contributions include: deriving closed-form solutions for weight quantization in each layer, significantly improving the representational capabilities of binarized weights; and developing a differentiable function for activation quantization, approximating the ideal multi-bit function while bypassing the extensive search for optimal settings. Through comprehensive experiments on CIFAR-10 and ImageNet datasets, we show that BWMA achieves notable accuracy improvements over existing methods, registering gains of 1.44\%-5.46\% and 0.35\%-5.37\% on respective datasets. Moreover, hardware simulation results indicate that 4-bit activation quantization strikes the optimal balance between hardware cost and model performance.