AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Fix your classifier: the marginal value of training the last weight layer

Elad Hoffer, Itay Hubara, Daniel Soudry

ICLR 2018 · Architectures and efficiency › RNNs and normalization · approach: Representation geometry and collapse · 12 citations (OpenAlex)

Read the paper

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Fix the classifier weights instead of learning themResNet56 · Cifar10 validationaccuracy: 93.03% -> 93.14%No clear effect
Fix the classifier weights instead of learning themDenseNet(k=12) · Cifar100 validationaccuracy: 77.73% -> 77.67%No clear effect
Fix the classifier weights instead of learning themResNet50 · ImageNet validationaccuracy: 75.3% -> 75.3%No clear effect
Fix the classifier weights instead of learning themDenseNet169 · ImageNet validationaccuracy: 76.2% -> 76%No clear effect
Fix the classifier weights instead of learning themShuffleNet · ImageNet validationaccuracy: 65.9% -> 65.4%Hurt
Replace learned embedding and classifier weights with fixed word2vec embeddings2-layer LSTM (h=512) · WikiText-2 validationperplexity: 74.1 -> 81.2Hurt
Use a random orthogonal transform instead of learned or pretrained word embeddings (ablation)Language model · WikiText-2perplexityHurt
Use fixed word2vec embeddings rather than a random orthogonal transform (ablation)Language model · WikiText-2perplexityImproved
Fix the softmax scale at α=1 or α=10 instead of learning it (ablation)ResNet56 · Cifar10 validationaccuracyNo clear effect
Replace softmax with the cosine angle loss and discard the scale parameter (ablation)Fixed-classifier models · Classification validationaccuracyHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy