Mixed-Precision Inference Quantization: Problem Resetting, Mapping math concept and Branch\&bound methods

Daning Cheng

Keywords: Mixed-Precision Quantization, Inference, Neural networks, Noise Robustness, NP hard

TL;DR: change mixed precision inference quantization problem into the problem which solved by branch and bound method

Abstract: Based on the model's resilience to computational noise, model quantization is important for compressing models and improving computing speed. Existing quantization techniques rely heavily on experience and "fine-tuning" skills. This study shows that in inference process, the mixed-precision quantization is a NP-hard problem and we designed a set of method to map mathematical method into pratical method. And our problem setting can be solved by branch and bound method with less computing resouces. We also show that how to set the quantization parameters in theorical method.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

Submission Guidelines: Yes

Please Choose The Closest Area That Your Submission Falls Into: General Machine Learning (ie none of the above)