NeurIPS 2024 accepted papers on Label-efficient and robust learning

Recognition and long-tail · 147 papers

• Depth Anything V2 (525 citations)
• Bench2Drive: Towards Multi-Ability Benchmarking of Closed-Loop End-To-End Autonomous Driving (81 citations)
• NAVSIM: Data-Driven Non-Reactive Autonomous Vehicle Simulation and Benchmarking (43 citations)
• A Closer Look at AUROC and AUPRC under Class Imbalance (34 citations)
• End-to-end Learnable Clustering for Intent Learning in Recommendation (18 citations)
• FreqBlender: Enhancing DeepFake Detection by Blending Frequency Knowledge (18 citations)
• DARNet: Dual Attention Refinement Network with Spatiotemporal Construction for Auditory Attention Detection (17 citations)
• SCOREQ: Speech Quality Assessment with Contrastive Regression (15 citations)
• FairMedFM: Fairness Benchmarking for Medical Imaging Foundation Models (14 citations)
• Interpreting CLIP with Sparse Linear Concept Embeddings (SpLiCE) (13 citations)
• Learning from Teaching Regularization: Generalizable Correlations Should be Easy to Imitate (12 citations)
• CARE: a Benchmark Suite for the Classification and Retrieval of Enzymes (10 citations)

and 135 more in the interactive map.

OOD and anomaly detection · 39 papers

• Unsupervised Anomaly Detection Algorithms on Real-world Data: How Many Do We Need? (12 citations)
• Kernel PCA for Out-of-Distribution Detection (10 citations)
• AdaNeg: Adaptive Negative Proxy Guided OOD Detection with Vision-Language Models (8 citations)
• Learning to Shape In-distribution Feature Space for Out-of-distribution Detection (8 citations)
• ResAD: A Simple Framework for Class Generalizable Anomaly Detection (7 citations)
• Conjugated Semantic Pool Improves OOD Detection with Pre-trained Vision-Language Models (5 citations)
• Credal Deep Ensembles for Uncertainty Quantification (5 citations)
• Hyper-opinion Evidential Deep Learning for Out-of-Distribution Detection (4 citations)
• Self-Calibrated Tuning of Vision-Language Models for Out-of-Distribution Detection (4 citations)
• What If the Input is Expanded in OOD Detection? (4 citations)
• Benchmarking Out-of-Distribution Generalization Capabilities of DNN-based Encoding Models for the Ventral Visual Cortex. (3 citations)
• One-to-Normal: Anomaly Personalization for Few-shot Anomaly Detection (3 citations)

and 27 more in the interactive map.

Self-supervised and contrastive · 37 papers

• EEGPT: Pretrained Transformer for Universal and Reliable Representation of EEG Signals (35 citations)
• Kuro Siwo: 33 billion $m^2$ under the water. A global multi-temporal satellite dataset for rapid flood mapping (14 citations)
• Connecting Joint-Embedding Predictive Architecture with Contrastive Self-supervised Learning (8 citations)
• Preventing Dimensional Collapse in Self-Supervised Learning via Orthogonality Regularization (8 citations)
• Your contrastive learning problem is secretly a distribution alignment problem (7 citations)
• Multi-Scale Representation Learning for Protein Fitness Prediction (5 citations)
• Self-Guided Masked Autoencoder (5 citations)
• SMART: Towards Pre-trained Missing-Aware Model for Patient Health Status Prediction (5 citations)
• Towards a "Universal Translator" for Neural Dynamics at Single-Cell, Single-Spike Resolution (5 citations)
• Return of Unconditional Generation: A Self-supervised Representation Generation Method (4 citations)
• Self-Labeling the Job Shop Scheduling Problem (4 citations)
• Unified Speech Recognition: A Single Model for Auditory, Visual, and Audiovisual Inputs (4 citations)

and 25 more in the interactive map.

Domain adaptation and generalization · 36 papers

• Learning Frequency-Adapted Vision Foundation Model for Domain Generalized Semantic Segmentation (25 citations)
• Attention Temperature Matters in ViT-Based Cross-Domain Few-Shot Learning (10 citations)
• A Closer Look at the CLS Token for Cross-Domain Few-Shot Learning (9 citations)
• Generalizable Person Re-identification via Balancing Alignment and Uniformity (9 citations)
• Lightweight Frequency Masker for Cross-Domain Few-Shot Semantic Segmentation (7 citations)
• Connectivity-Driven Pseudo-Labeling Makes Stronger Cross-Domain Segmenters (6 citations)
• Enhancing Domain Adaptation through Prompt Gradient Alignment (6 citations)
• Style Adaptation and Uncertainty Estimation for Multi-Source Blended-Target Domain Adaptation (6 citations)
• Geodesic Optimization for Predictive Shift Adaptation on EEG data (5 citations)
• Samba: Severity-aware Recurrent Modeling for Cross-domain Medical Image Grading (5 citations)
• Federated Graph Learning for Cross-Domain Recommendation (4 citations)
• IODA: Instance-Guided One-shot Domain Adaptation for Super-Resolution (3 citations)

and 24 more in the interactive map.

Test-time adaptation · 20 papers

• Cross-Device Collaborative Test-Time Adaptation (8 citations)
• BoostAdapter: Improving Vision-Language Test-Time Adaptation via Regional Bootstrapping (6 citations)
• Historical Test-time Prompt Tuning for Vision Foundation Models (6 citations)
• L-TTA: Lightweight Test-Time Adaptation Using a Versatile Stem Layer (6 citations)
• Test-Time Dynamic Image Fusion (6 citations)
• A Data-Centric Perspective on Evaluating Machine Learning Models for Tabular Data (5 citations)
• EMGBench: Benchmarking Out-of-Distribution Generalization and Adaptation for Electromyography (4 citations)
• Protected Test-Time Adaptation via Online Entropy Matching: A Betting Approach (4 citations)
• Persistent Test-time Adaptation in Recurring Testing Scenarios (3 citations)
• Dual-frame Fluid Motion Estimation with Test-time Optimization and Zero-divergence Loss (2 citations)
• Reconstruct and Match: Out-of-Distribution Robustness via Topological Homogeneity (2 citations)
• Test-time Adaptation in Non-stationary Environments via Adaptive Representation Alignment (2 citations)

and 8 more in the interactive map.

Noisy labels · 18 papers

• Imprecise Label Learning: A Unified Framework for Learning with Various Imprecise Label Configurations (8 citations)
• Curriculum Fine-tuning of Vision Foundation Model for Medical Image Classification Under Label Noise (5 citations)
• NoiseGPT: Label Noise Detection and Rectification through Probability Curvature (5 citations)
• Collaborative Refining for Learning from Inaccurate Labels (3 citations)
• Stochastic Amortization: A Unified Approach to Accelerate Feature and Data Attribution (2 citations)
• CoSW: Conditional Sample Weighting for Smoke Segmentation with Label Noise (1 citation)
• dopanim: A Dataset of Doppelganger Animals with Noisy Annotations from Multiple Humans (1 citation)
• Information-theoretic Limits of Online Classification with Noisy Labels (1 citation)
• IWBVT: Instance Weighting-based Bias-Variance Trade-off for Crowdsourcing (1 citation)
• KFNN: K-Free Nearest Neighbor For Crowdsourcing (1 citation)
• Noisy Label Learning with Instance-Dependent Outliers: Identifiability via Crowd Wisdom (1 citation)
• Noisy Ostracods: A Fine-Grained, Imbalanced Real-World Dataset for Benchmarking Robust Machine Learning and Label Correction Methods (1 citation)

and 6 more in the interactive map.

Zero-/few-shot and meta-learning · 16 papers

• Pretrained Optimization Model for Zero-Shot Black Box Optimization (5 citations)
• Meta-Exploiting Frequency Prior for Cross-Domain Few-Shot Learning (4 citations)
• D2R2: Diffusion-based Representation with Random Distance Matching for Tabular Few-shot Learning (3 citations)
• OTTER: Effortless Label Distribution Adaptation of Zero-shot Models (3 citations)
• Dual Risk Minimization: Towards Next-Level Robustness in Fine-tuning Zero-Shot Models (2 citations)
• Be Confident in What You Know: Bayesian Parameter Efficient Fine-Tuning of Vision Foundation Models (1 citation)
• EEVR: A Dataset of Paired Physiological Signals and Textual Descriptions for Joint Emotion Representation Learning (1 citation)
• Few-shot Algorithms for Consistent Neural Decoding (FALCON) Benchmark (1 citation)
• How Molecules Impact Cells: Unlocking Contrastive PhenoMolecular Retrieval (1 citation)
• Listenable Maps for Zero-Shot Audio Classifiers (1 citation)
• Robust Fine-tuning of Zero-shot Models via Variance Reduction (1 citation)
• Class Distribution Shifts in Zero-Shot Learning: Learning Robust Representations

and 4 more in the interactive map.

Semi-supervised learning · 13 papers

• Real-world Image Dehazing with Coherence-based Pseudo Labeling and Cooperative Unfolding Network (17 citations)
• Boosting Semi-Supervised Scene Text Recognition via Viewing and Summarizing (3 citations)
• OwMatch: Conditional Self-Labeling with Consistency for Open-World Semi-Supervised Learning (3 citations)
• RankUp: Boosting Semi-Supervised Regression with an Auxiliary Ranking Classifier (2 citations)
• Selective Generation for Controllable Language Models (2 citations)
• Semi-supervised Knowledge Transfer Across Multi-omic Single-cell Data (2 citations)
• Semi-Supervised Sparse Gaussian Classification: Provable Benefits of Unlabeled Data (2 citations)
• Data Augmentation with Diffusion for Open-Set Semi-Supervised Learning (1 citation)
• Reinforcement Learning Guided Semi-Supervised Learning (1 citation)
• Semi-supervised Multi-label Learning with Balanced Binary Angular Margin Loss (1 citation)
• Continuous Contrastive Learning for Long-Tailed Semi-Supervised Recognition
• Enhancing Semi-Supervised Learning via Representative and Diverse Sample Selection

and 1 more in the interactive map.

Knowledge distillation · 12 papers

• Decoupled Kullback-Leibler Divergence Loss (41 citations)
• Wasserstein Distance Rivals Kullback-Leibler Divergence for Knowledge Distillation (21 citations)
• LaKD: Length-agnostic Knowledge Distillation for Trajectory Prediction with Any Length Observations (8 citations)
• Computerized Adaptive Testing via Collaborative Ranking (3 citations)
• Flipped Classroom: Aligning Teacher Attention with Student in Generalized Category Discovery (3 citations)
• UDON: Universal Dynamic Online distillatioN for generic image representations (2 citations)
• Understanding the Gains from Repeated Self-Distillation (2 citations)
• Induced Model Matching: Restricted Models Help Train Full-Featured Models (1 citation)
• Reinforced Cross-Domain Knowledge Distillation on Time Series Data (1 citation)
• A Label is Worth A Thousand Images in Dataset Distillation
• Are Large-scale Soft Labels Necessary for Large-scale Dataset Distillation?
• CIFD: Controlled Information Flow to Enhance Knowledge Distillation