Revisiting the Transferability of Supervised Pretraining: an MLP Perspective

Abstract

The pretrain-finetune paradigm is a classical pipeline in visual learning. Recent progress on unsupervised pretraining methods shows superior transfer performance to their supervised counterparts. This paper revisits this phenomenon and sheds new light on understanding the transferability gap between unsupervised and supervised pretraining from a multilayer perceptron (MLP) perspective. While previous works focus on the effectiveness of MLP on unsupervised image classification where pretraining and evaluation are conducted on the same dataset, we reveal that the MLP projector is also the key factor to better transferability of unsupervised pretraining methods than supervised pretraining methods. Based on this observation, we attempt to close the transferability gap between supervised and unsupervised pretraining by adding an MLP projector before the classifier in supervised pretraining. Our analysis indicates that the MLP projector can help retain intra-class variation of visual features, decrease the feature distribution distance between pretraining and evaluation datasets, and reduce feature redundancy. Extensive experiments on public benchmarks demonstrate that the added MLP projector significantly boosts the transferability of supervised pretraining, e.g. +7.2% top-1 accuracy on the concept generalization task, +5.8% top-1 accuracy for linear evaluation on 12-domain classification tasks, and +0.8% AP on COCO object detection task, making supervised pretraining comparable or even better than unsupervised pretraining.

arXiv commentAccepted by CVPR 2022. [camera ready with supplement]

References in corpus (13)

• Bootstrap your own latent: A new approach to self-supervised Learning
• Improved Baselines with Momentum Contrastive Learning
• Unsupervised Learning of Visual Features by Contrasting Cluster Assignments
• Skin Lesion Analysis Toward Melanoma Detection 2018: A Challenge Hosted by the International Skin Imaging Collaboration (ISIC)
• Barlow Twins: Self-Supervised Learning via Redundancy Reduction
• DeepWeeds: A Multiclass Weed Species Image Dataset for Deep Learning
• Large Batch Training of Convolutional Networks
• Big Self-Supervised Models are Strong Semi-Supervised Learners
• Supervised Contrastive Learning
• A critical analysis of self-supervision, or what we can learn from a single image
• What makes instance discrimination good for transfer learning?
• KaoKore: A Pre-modern Japanese Art Facial Expression Dataset
• Towards Understanding the Transferability of Deep Representations