Paper Image

Optimizing Deep Shift Networks for Accuracy and Efficiency

Paper Authors:

Leona Hennig,

Tanja Tornede,

Marius Lindauer

Bullets

Key Details

Employs automated ML to optimize deep shift networks

Combines multi-objective and multi-fidelity optimization

Tunes models for high accuracy and low energy use

Achieves over 80% accuracy with low emissions

AI generated summary

Optimizing Deep Shift Networks for Accuracy and Efficiency

This paper proposes an automated machine learning approach to optimize deep shift neural networks, which replace costly multiplication operations with more efficient bit shift operations. By combining multi-objective and multi-fidelity hyperparameter optimization, models are tuned to maximize accuracy while minimizing energy consumption. Over 80% test accuracy is achieved with low emissions.

Answers from this paper

Comments

No comments yet, be the first to start the conversation...

Sign up to comment on this paper

Sign Up