Learning Compact Features via In-Training Representation Alignment

Deep neural networks (DNNs) for supervised learning can be viewed as a pipeline of the feature extractor (i.e., last hidden layer) and a linear classifier (i.e., output layer) that are trained jointly with stochastic gradient descent (SGD) on the loss function (e.g., cross-entropy). In each epoch, the true gradient of the loss function is estimated using a mini-batch sampled from the training set and model parameters are then updated with the mini-batch gradients. Although the latter provides an unbiased estimation of the former, they are subject to substantial variances derived from the size and number of sampled mini-batches, leading to noisy and jumpy updates. To stabilize such undesirable variance in estimating the true gradients, we propose In-Training Representation Alignment (ITRA) that explicitly aligns feature distributions of two different mini-batches with a matching loss in the SGD training process. We also provide a rigorous analysis of the desirable effects of the matching loss on feature representation learning: (1) extracting compact feature representation; (2) reducing over-adaption on mini-batches via an adaptive weighting mechanism; and (3) accommodating to multi-modalities. Finally, we conduct large-scale experiments on both image and text classifications to demonstrate its superior performance to the strong baselines.
Comment: 11 pages, 4 figures, 6 tables. Accepted for publication by AAAI-23. arXiv admin note: text overlap with arXiv:2002.09917