pseudo-labels
Pseudo-labels are labels generated from the model’s own predictions on unlabelled data. They are often used in semi-supervised learning strategies to increase the effective size of training data, by treating high-confidence predictions as if they were true labels.
- Can Class-Priors Help Single-Positive Multi-Label Learning?
- Diffusion-Driven Progressive Target Manipulation for Source-Free Domain Adaptation
- Disentangling Latent Shifts of In-Context Learning with Weak Supervision
- Dual Prototype-Enhanced Contrastive Framework for Class-Imbalanced Graph Domain Adaptation
- FrameShield: Adversarially Robust Video Anomaly Detection
- L2DGCN: Learnable Enhancement and Label Selection Dynamic Graph Convolutional Networks for Mitigating Degree Bias
- LVLM-Driven Attribute-Aware Modeling for Visible-Infrared Person Re-Identification
- Mint: A Simple Test-Time Adaptation of Vision-Language Models against Common Corruptions
- Prediction-Powered Semi-Supervised Learning with Online Power Tuning
- Revisiting Semi-Supervised Learning in the Era of Foundation Models
- Self Iterative Label Refinement via Robust Unlabeled Learning
- Self-Training with Dynamic Weighting for Robust Gradual Domain Adaptation
- Spatiotemporal Consensus with Scene Prior for Unsupervised Domain Adaptive Person Search
- TRUST: Test-Time Refinement using Uncertainty-Guided SSM Traverses
- Uncertainty-Informed Meta Pseudo Labeling for Surrogate Modeling with Limited Labeled Data