Showing cs.LGShow all
2 papers · 1 filter
cs.LG2026
Layerwise Progressive Freezing: A Training Scaffold for Depth-Scalable Binary Networks
Evan Gibson Smith, Bashima Islam
Training binary neural networks (BNNs) from scratch is dominated by the straight-through estimator (STE), whose forward/backward mismatch produces severe accuracy degradation as ne…
cs.LG2026
Layerwise Progressive Freezing Enables STE-Free Training of Deep Binary Neural Networks
Evan Gibson Smith, Bashima Islam
We investigate progressive freezing as an alternative to straight-through estimators (STE) for training binary networks from scratch. Under controlled training conditions, we find…