Evan Gibson Smith

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 networks deepen. We study an orthogonal axis: when and where binarization is enforced during training.…Read more