Delving into Pre-training for Domain Transfer: A Broad Study of Pre-training for Domain Generalization and Domain Adaptation

Keyphrases

• Domain Generalization 100%
• Domain Adaptation 100%
• Domain Transfer 100%
• Pre-training Datasets 37%
• Transfer Method 25%
• Alignment Algorithm 12%
• Adaptation Techniques 12%
• Network Architecture 12%
• Overfitting 12%
• Robust Features 12%
• Training Approaches 12%
• Data Leakage 12%
• Impact Performance 12%
• Feature Representation 12%
• Adaptive Algorithm 12%
• Object Detection 12%
• Large-scale Data 12%
• Class number 12%
• Semantic Segmentation 12%
• Object Semantics 12%
• ImageNet 12%
• Unsupervised Domain Adaptation 12%
• Deep Learning Model 12%
• Downstream Task 12%
• Multi-source Domain Adaptation 12%
• Domain Shift 12%
• Web-scale 12%
• Performance Gains 12%
• ResNet 12%
• Reliable Features 12%
• Generalization Techniques 12%
• Transfer Studies 12%
• New Adaptation 12%
• Pre-training Objectives 12%
• Universal Domain Adaptation (UniDA) 12%
• ImageNet-1K 12%
• Unseen Domain 12%

Computer Science

• Domain Adaptation 100%
• Training Dataset 60%
• Vertebral Column 40%
• Performance Gain 20%
• Network Architecture 20%
• Image Segmentation 20%
• Object Detection 20%
• Large Scale Data 20%
• Unsupervised Domain Adaptation 20%
• Deep Learning Model 20%
• Residual Neural Network 20%
• Training Objective 20%
• Universal Domain 20%
• Source-Free Domain Adaptation 20%

Social Sciences

• Domain Generalization 100%
• Resnet 33%
• Deep Learning Method 33%
• Object Detection 33%
• Image Segmentation 33%
• Large Scale Data 33%