Hallucination Improves the Performance of Unsupervised Visual Representation Learning

Keyphrases

• Hallucinations 100%
• Visual Learning 100%
• Contrastive Learning 100%
• Learning Model 75%
• ImageNet 50%
• Mutual Information 25%
• Feature Space 25%
• Encoder 25%
• Overfitting 25%
• Feature Embedding 25%
• Positive Samples 25%
• Object Detection 25%
• Transformation Invariant 25%
• Self-supervised Learning 25%
• Object Segmentation 25%
• Invariant Features 25%
• Nonlinear Operation 25%
• Downstream Task 25%
• Positive Pairs 25%
• CIFAR-10 25%
• Linear Classification 25%
• Pre-training Tasks 25%
• Semantic Contrast 25%
• Siamese Structure 25%
• SimSiam 25%
• SimCLR 25%

Computer Science

• Representation Learning 100%
• Visual Representation 100%
• Contrastive Learning 100%
• Mutual Information 25%
• Feature Space 25%
• Object Detection 25%
• Sufficient Number 25%
• Self-Supervised Learning 25%
• Coder 25%