Learning Multimodal Fixed-Point Weights using Gradient Descent
Machine Learning
2019-07-18 v1 Computer Vision and Pattern Recognition
Abstract
Due to their high computational complexity, deep neural networks are still limited to powerful processing units. To promote a reduced model complexity by dint of low-bit fixed-point quantization, we propose a gradient-based optimization strategy to generate a symmetric mixture of Gaussian modes (SGM) where each mode belongs to a particular quantization stage. We achieve 2-bit state-of-the-art performance and illustrate the model's ability for self-dependent weight adaptation during training.
Cite
@article{arxiv.1907.07220,
title = {Learning Multimodal Fixed-Point Weights using Gradient Descent},
author = {Lukas Enderich and Fabian Timm and Lars Rosenbaum and Wolfram Burgard},
journal= {arXiv preprint arXiv:1907.07220},
year = {2019}
}
Comments
presented at ESANN 2019 (European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning)