SEED: Self-supervised Distillation For Visual Representation

2021/01/12 by Fang, Zhiyuan, Wang, Jianfeng, Wang, Lijuan +3 · 7 citations
#Artificial Intelligence (cs.AI) #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences

paper · doi:10.48550/arxiv.2101.04731

Abstract

This paper is concerned with self-supervised learning for small models. The problem is motivated by our empirical studies that while the widely used contrastive self-supervised learning method has shown great progress on large model training, it does not work well for small models. To address this problem, we propose a new learning paradigm, named SElf-SupErvised Distillation (SEED), where we leverage a larger network (as Teacher) to transfer its representational knowledge into a smaller architecture (as Student) in a self-supervised fashion. Instead of directly learning from unlabeled data, we train a student encoder to mimic the similarity score distribution inferred by a teacher over a set of instances. We show that SEED dramatically boosts the performance of small networks on downstream tasks. Compared with self-supervised baselines, SEED improves the top-1 accuracy from 42.2% to 67.6% on EfficientNet-B0 and from 36.3% to 68.2% on MobileNet-v3-Large on the ImageNet-1k dataset.

Cited by

• PSMamba: Progressive Self-supervised Vision Mamba for Plant Disease Recognition
• Super4DR: 4D Radar-centric Self-supervised Odometry and Gaussian-based Map Optimization
• When Better Teachers Don't Make Better Students: Revisiting Knowledge Distillation for CLIP Models in VQA
• Distilling Multilingual Vision-Language Models: When Smaller Models Stay Multilingual
• Learning Task-Agnostic Representations through Multi-Teacher Distillation
• TTRV: Test-Time Reinforcement Learning for Vision Language Models
• HanjaBridge: Resolving Semantic Ambiguity in Korean LLMs via Hanja-Augmented Pre-Training

Related

• Beyond Self-Supervision: A Simple Yet Effective Network Distillation Alternative to Improve Backbones
• A Simple Recipe for Competitive Low-compute Self supervised Vision Models
• DMT: Comprehensive Distillation with Multiple Self-supervised Teachers
• Distill on the Go: Online knowledge distillation in self-supervised learning
• Boost Self-Supervised Dataset Distillation via Parameterization, Predefined Augmentation, and Approximation