AAAI 2022 accepted papers on Label-efficient and robust learning
200 of the 1,319 papers accepted at AAAI 2022 (15.2%) are on Label-efficient and robust learning, compared with 15.8% at AAAI 2021. They fall into 9 topics.
See AAAI 2022 in the interactive map
Working on something in this area? Describe your idea in scime atlas to see the closest papers by meaning and how crowded the spot has become.
Topics and papers
Recognition and long-tail · 98 papers
- Unified Named Entity Recognition as Word-Word Relation Classification (392 citations)
- Dual Contrastive Learning for General Face Forgery Detection (199 citations)
- Exploiting Fine-Grained Face Forgery Clues via Progressive Enhancement Learning (157 citations)
- Contrast and Generation Make BART a Good Dialogue Emotion Recognizer (122 citations)
- Modality-Adaptive Mixup and Invariant Decomposition for RGB-Infrared Person Re-identification (117 citations)
- Multi-View Intent Disentangle Graph Networks for Bundle Recommendation (108 citations)
- Invariant Information Bottleneck for Domain Generalization (102 citations)
- Do Feature Attribution Methods Correctly Attribute Features? (94 citations)
- BERTMap: A BERT-Based Ontology Alignment System (81 citations)
- Local Similarity Pattern and Cost Self-Reassembling for Deep Stereo Matching Networks (78 citations)
- Sequence Level Contrastive Learning for Text Summarization (78 citations)
- Explainable Metaphor Identification Inspired by Conceptual Metaphor Theory (73 citations)
and 86 more in the interactive map.
Domain adaptation and generalization · 24 papers
- Active Learning for Domain Adaptation: An Energy-Based Approach (135 citations)
- PureGaze: Purifying Gaze Feature for Generalizable Gaze Estimation (97 citations)
- SECRET: Self-Consistent Pseudo Label Refinement for Unsupervised Domain Adaptive Person Re-identification (94 citations)
- Delving into Probabilistic Uncertainty for Unsupervised Domain Adaptive Person Re-identification (71 citations)
- DIRL: Domain-Invariant Representation Learning for Generalizable Semantic Segmentation (61 citations)
- Denoised Maximum Classifier Discrepancy for Source-Free Unsupervised Domain Adaptation (58 citations)
- Reliability Exploration with Self-Ensemble Learning for Domain Adaptive Person Re-identification (48 citations)
- Unpaired Multi-Domain Stain Transfer for Kidney Histopathological Images (47 citations)
- Evidential Neighborhood Contrastive Learning for Universal Domain Adaptation (41 citations)
- MultiplexNet: Towards Fully Satisfied Logical Constraints in Neural Networks (33 citations)
- Debiased Batch Normalization via Gaussian Process for Generalizable Person Re-identification (27 citations)
- Style Mixing and Patchwise Prototypical Matching for One-Shot Unsupervised Domain Adaptive Semantic Segmentation (27 citations)
and 12 more in the interactive map.
Zero-/few-shot and meta-learning · 19 papers
- ContrastNet: A Contrastive Learning Framework for Few-Shot Text Classification (103 citations)
- Semantic Feature Extraction for Generalized Zero-Shot Learning (53 citations)
- Inductive Relation Prediction by BERT (49 citations)
- Adaptive Poincaré Point to Set Distance for Few-Shot Classification (42 citations)
- DMN4: Few-Shot Learning via Discriminative Mutual Nearest Neighbor Neural Network (39 citations)
- Dual Attention Networks for Few-Shot Fine-Grained Recognition (39 citations)
- Label Hallucination for Few-Shot Classification (36 citations)
- Zero-Shot Audio Source Separation through Query-Based Learning from Weakly-Labeled Data (31 citations)
- Boosting Generative Zero-Shot Learning by Synthesizing Diverse Features with Attribute Augmentation (25 citations)
- Domain Disentangled Generative Adversarial Network for Zero-Shot Sketch-Based 3D Shape Retrieval (25 citations)
- ALP: Data Augmentation Using Lexicalized PCFGs for Few-Shot Text Classification (24 citations)
- Powering Finetuning in Few-Shot Learning: Domain-Agnostic Bias Reduction with Selected Sampling (21 citations)
and 7 more in the interactive map.
Self-supervised and contrastive · 17 papers
- SSAST: Self-Supervised Audio Spectrogram Transformer (281 citations)
- Contrastive Learning from Extremely Augmented Skeleton Sequences for Self-Supervised Action Recognition (180 citations)
- TransMEF: A Transformer-Based Multi-Exposure Image Fusion Framework Using Self-Supervised Multi-Task Learning (179 citations)
- MODNet: Real-Time Trimap-Free Portrait Matting via Objective Decomposition (156 citations)
- Mutual Contrastive Learning for Visual Representation Learning (84 citations)
- Intra-Inter Subject Self-Supervised Learning for Multivariate Cardiac Signals (51 citations)
- Perceiving Stroke-Semantic Context: Hierarchical Contrastive Learning for Robust Scene Text Recognition (46 citations)
- Self-Supervised Pre-training for Protein Embeddings Using Tertiary Structures (29 citations)
- Can Semantic Labels Assist Self-Supervised Visual Representation Learning? (22 citations)
- InsCLR: Improving Instance Retrieval with Self-Supervision (14 citations)
- Task-Customized Self-Supervised Pre-training with Scalable Dynamic Routing (10 citations)
- On the Efficacy of Small Self-Supervised Contrastive Models without Distillation Signals (8 citations)
and 5 more in the interactive map.
Knowledge distillation · 13 papers
- Knowledge Distillation for Object Detection via Rank Mimicking and Prediction-Guided Feature Imitation (97 citations)
- ADD: Frequency Attention and Multi-View Based Knowledge Distillation to Detect Low-Quality Compressed Deepfake Images (61 citations)
- Robust and Resource-Efficient Data-Free Knowledge Distillation by Generative Pseudo Replay (38 citations)
- Cross-Task Knowledge Distillation in Multi-Task Recommendation (37 citations)
- LGD: Label-Guided Self-Distillation for Object Detection (35 citations)
- Improving Neural Cross-Lingual Abstractive Summarization via Employing Optimal Transport Distance for Knowledge Distillation (33 citations)
- Cosine Model Watermarking against Ensemble Distillation (27 citations)
- No Task Left Behind: Multi-Task Learning of Knowledge Tracing and Option Tracing for Better Student Assessment (17 citations)
- Boosting Contrastive Learning with Relation Knowledge Distillation (12 citations)
- Safe Distillation Box (11 citations)
- TrustAL: Trustworthy Active Learning Using Knowledge Distillation (9 citations)
- Contact-Distil: Boosting Low Homologous Protein Contact Map Prediction by Self-Supervised Distillation (4 citations)
and 1 more in the interactive map.
Semi-supervised learning · 12 papers
- Enhancing Pseudo Label Quality for Semi-supervised Domain-Generalized Medical Image Segmentation (114 citations)
- GALAXY: A Generative Pre-trained Model for Task-Oriented Dialog with Semi-supervised Learning and Explicit Policy Injection (71 citations)
- LaSSL: Label-Guided Self-Training for Semi-supervised Learning (45 citations)
- GuidedMix-Net: Semi-supervised Semantic Segmentation by Using Labeled Images as Reference (28 citations)
- Barely-Supervised Learning: Semi-supervised Learning with Very Few Labeled Images (26 citations)
- Semi-supervised Learning with Multi-Head Co-Training (24 citations)
- Not All Parameters Should Be Treated Equally: Deep Safe Semi-supervised Learning under Class Distribution Mismatch (23 citations)
- Contrast-Enhanced Semi-supervised Text Classification with Few Labels (20 citations)
- Ensemble Semi-supervised Entity Alignment via Cycle-Teaching (17 citations)
- SJDL-Vehicle: Semi-supervised Joint Defogging Learning for Foggy Vehicle Re-identification (17 citations)
- ASM2TV: An Adaptive Semi-supervised Multi-Task Multi-View Learning Framework for Human Activity Recognition (11 citations)
- A Semi-supervised Learning Approach with Two Teachers to Improve Breakdown Identification in Dialogues (4 citations)
OOD and anomaly detection · 10 papers
- Deep One-Class Classification via Interpolated Gaussian Descriptor (116 citations)
- Zero-Shot Out-of-Distribution Detection Based on the Pre-trained Model CLIP (116 citations)
- Provable Guarantees for Understanding Out-of-Distribution Detection (52 citations)
- On the Impact of Spurious Correlation for Out-of-Distribution Detection (31 citations)
- Out of Distribution Data Detection Using Dropout Bayesian Neural Networks (27 citations)
- iDECODe: In-Distribution Equivariance for Conformal Out-of-Distribution Detection (20 citations)
- Transferring the Contamination Factor between Anomaly Detection Domains by Shape Similarity (14 citations)
- OoDHDR-Codec: Out-of-Distribution Generalization for HDR Image Compression (13 citations)
- Gradient-Based Novelty Detection Boosted by Self-Supervised Binary Classification (12 citations)
- Exploiting Mixed Unlabeled Data for Detecting Samples of Seen and Unseen Out-of-Distribution Classes (4 citations)
Noisy labels · 4 papers
- Uncertainty-Aware Learning against Label Noise on Imbalanced Datasets (54 citations)
- Delving into Sample Loss Curve to Embrace Noisy and Imbalanced Data (37 citations)
- Noise-Robust Learning from Multiple Unsupervised Sources of Inferred Labels (12 citations)
- Deep Neural Networks Learn Meta-Structures from Noisy Labels in Semantic Segmentation (11 citations)
Test-time adaptation · 3 papers
- DDG-DA: Data Distribution Generation for Predictable Concept Drift Adaptation (52 citations)
- Single-Domain Generalization in Medical Image Segmentation via Test-Time Adaptation from Shape Dictionary (44 citations)
- Generalizable Person Re-identification via Self-Supervised Batch Norm Test-Time Adaption (19 citations)
