Cursor-based adaptive quantization for deep neural networks
US12039427B2 · kind B2 · utility
Assignees
Inventors
Key dates
| Filing date | Sep 24, 2019 |
| Grant date | Jul 16, 2024 |
| Priority date | — |
| Expiry date | Jun 1, 2042 |
Classification
- Technology area (CPC G)Physics
- CPC primaryG06F16/28
- WIPO fieldComputer technology
- WIPO sectorElectrical engineering
Abstract
Deep neural networks (DNN) model quantization may be used to reduce storage and computation burdens by decreasing the bit width. Presented herein are novel cursor-based adaptive quantization embodiments. In embodiments, a multiple bits quantization mechanism is formulated as a differentiable architecture search (DAS) process with a continuous cursor that represents a possible quantization bit. In embodiments, the cursor-based DAS adaptively searches for a quantization bit for each layer. The DAS process may be accelerated via an alternative approximate optimization process, which is designed for mixed quantization scheme of a DNN model. In embodiments, a new loss function is used in the search process to simultaneously optimize accuracy and parameter size of the model. In a quantization step, the closest two integers to the cursor may be adopted as the bits to quantize the DNN together to reduce the quantization noise and avoid the local convergence problem.
Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.