Regularizing Neural Networks by Penalizing Confident Output Distributions
About
We systematically explore regularizing neural networks by penalizing low entropy output distributions. We show that penalizing low entropy output distributions, which has been shown to improve exploration in reinforcement learning, acts as a strong regularizer in supervised learning. Furthermore, we connect a maximum entropy based confidence penalty to label smoothing through the direction of the KL divergence. We exhaustively evaluate the proposed confidence penalty and label smoothing on 6 common benchmarks: image classification (MNIST and Cifar-10), language modeling (Penn Treebank), machine translation (WMT'14 English-to-German), and speech recognition (TIMIT and WSJ). We find that both label smoothing and the confidence penalty improve state-of-the-art models across benchmarks without modifying existing hyperparameters, suggesting the wide applicability of these regularizers.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Semantic segmentation | PASCAL VOC 2012 (val) | Mean IoU71.16 | 2040 | |
| Fine-grained Image Classification | CUB200 2011 (test) | Accuracy73.51 | 536 | |
| Fine-grained Image Classification | Stanford Cars (test) | Accuracy73.78 | 348 | |
| Image Classification | ImageNet LT | Top-1 Accuracy37.69 | 251 | |
| Machine Translation | IWSLT De-En 2014 (test) | BLEU34.2 | 146 | |
| Fine-grained Image Classification | Stanford Dogs (test) | Accuracy74.41 | 117 | |
| Out-of-Distribution Detection | CIFAR-10 vs CIFAR-100 (test) | -- | 93 | |
| Machine Translation | IWSLT En-De 2014 (test) | BLEU27.9 | 92 | |
| Text Classification | 20 Newsgroups (test) | Accuracy66.48 | 71 | |
| Object Detection | PASCAL VOC to Water Color (test) | mAP39.4 | 64 |