| Title | Quantized Neural Networks: Characterization and Holistic Optimization |
| Publication Type | Conference Paper |
| Year of Publication | 2020 |
| Authors | Boo, Yoonho, Shin, Sungho, Sung, Wonyong |
| Conference Name | 2020 IEEE Workshop on Signal Processing Systems (SiPS) |
| Date Published | oct |
| Keywords | activation quantization, holistic approach, neural network resiliency, Neural networks, Optimization methods, pubcrawl, Quantization (signal), quantized deep neural network, resilience, Resiliency, Sensitivity, Throughput, Training, weight quantization |
| Abstract | Quantized deep neural networks (QDNNs) are necessary for low-power, high throughput, and embedded applications. Previous studies mostly focused on developing optimization methods for the quantization of given models. However, quantization sensitivity depends on the model architecture. Also, the characteristics of weight and activation quantization are quite different. This study proposes a holistic approach for the optimization of QDNNs, which contains QDNN training methods as well as quantization-friendly architecture design. Synthesized data is used to visualize the effects of weight and activation quantization. The results indicate that deeper models are more prone to activation quantization, while wider models improve the resiliency to both weight and activation quantization. |
| DOI | 10.1109/SiPS50750.2020.9195245 |
| Citation Key | boo_quantized_2020 |
Quantized Neural Networks: Characterization and Holistic Optimization