Learning multimodal fixed-point weights using gradient descent.
2019
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.
- Correction
- Source
- Cite
- Save
- Machine Reading By IdeaReader
0
References
10
Citations
NaN
KQI