A Novel Low-Bit Quantization Strategy for Compressing Deep Neural Networks

<div>Convolution operation pipeline. (a) General convolution operation without quantization of weight and activation. (b) Description of proposed method with weight and activation quantized by low-bit.</div>

Computational Intelligence and Neuroscience

fig1

Figure 1

Figure 1: A Novel Low-Bit Quantization Strategy for Compressing Deep Neural Networks 