ICML 2024 accepted papers on Label-efficient and robust learning

Recognition and long-tail · 112 papers

• Position: The No Free Lunch Theorem, Kolmogorov Complexity, and the Role of Inductive Biases in Machine Learning (15 citations)
• CLIPZyme: Reaction-Conditioned Virtual Screening of Enzymes (9 citations)
• How Learning by Reconstruction Produces Uninformative Features For Perception (5 citations)
• Masked Face Recognition with Generative-to-Discriminative Representations (5 citations)
• Don't trust your eyes: on the (un)reliability of feature visualizations (4 citations)
• Estimating Canopy Height at Scale (4 citations)
• Robustness of Deep Learning for Accelerated MRI: Benefits of Diverse Training Data (4 citations)
• $f$-Divergence Based Classification: Beyond the Use of Cross-Entropy (3 citations)
• Contrastive Learning for Clinical Outcome Prediction with Partial Data Sources (3 citations)
• FlashST: A Simple and Universal Prompt-Tuning Framework for Traffic Prediction (3 citations)
• Improving Antibody Humanness Prediction using Patent Data (3 citations)
• Learning to Intervene on Concept Bottlenecks (3 citations)

and 100 more in the interactive map.

Self-supervised and contrastive · 21 papers

• Auto-Linear Phenomenon in Subsurface Imaging (4 citations)
• Binning as a Pretext Task: Improving Self-Supervised Learning in Tabular Domains (1 citation)
• Improved Generalization of Weight Space Networks via Augmentations (1 citation)
• Information Flow in Self-Supervised Learning (1 citation)
• Language Models as Semantic Indexers (1 citation)
• Matrix Information Theory for Self-Supervised Learning (1 citation)
• Self-Supervised Interpretable End-to-End Learning via Latent Functional Modularity (1 citation)
• Enhancing Trajectory Prediction through Self-Supervised Waypoint Distortion Prediction
• Knowledge Transfer from Vision Foundation Models for Efficient Training of Small Task-specific Models
• Look Ahead or Look Around? A Theoretical Comparison Between Autoregressive and Masked Pretraining
• On the Effectiveness of Supervision in Asymmetric Non-Contrastive Learning
• Regularizing with Pseudo-Negatives for Continual Self-Supervised Learning

and 9 more in the interactive map.

Domain adaptation and generalization · 20 papers

• A Circuit Domain Generalization Framework for Efficient Logic Synthesis in Chip Design (3 citations)
• DOGE: Domain Reweighting with Generalization Estimation (2 citations)
• Exploring Training on Heterogeneous Data with Mixture of Low-rank Adapters (1 citation)
• Few-shot Adaptation to Distribution Shifts By Mixing Source and Target Embeddings (1 citation)
• Hyperbolic Active Learning for Semantic Segmentation under Domain Shift (1 citation)
• Bayesian Power Steering: An Effective Approach for Domain Adaptation of Diffusion Models
• Connect Later: Improving Fine-tuning for Robustness with Targeted Augmentations
• Contrastive Representation for Data Filtering in Cross-Domain Offline Reinforcement Learning
• Cross-domain Open-world Discovery
• EvoluNet: Advancing Dynamic Non-IID Transfer Learning on Graphs
• IW-GAE: Importance weighted group accuracy estimation for improved calibration and model selection in unsupervised domain adaptation
• Out of the Ordinary: Spectrally Adapting Regression for Covariate Shift

and 8 more in the interactive map.

OOD and anomaly detection · 17 papers

• CRoFT: Robust Fine-Tuning with Concurrent Optimization for OOD Generalization and Open-Set OOD Detection (3 citations)
• Uncertainty Estimation by Density Aware Evidential Deep Learning (2 citations)
• Beyond Individual Input for Deep Anomaly Detection on Tabular Data (1 citation)
• Domain Generalisation via Imprecise Learning (1 citation)
• Envisioning Outlier Exposure by Large Language Models for Out-of-Distribution Detection (1 citation)
• Fast Decision Boundary based Out-of-Distribution Detector (1 citation)
• Contamination-Resilient Anomaly Detection via Adversarial Learning on Partially-Observed Normal and Anomalous Data
• Discovering Environments with XRM
• Online Isolation Forest
• Out-of-Distribution Detection via Deep Multi-Comprehension Ensemble
• Split-Ensemble: Efficient OOD-aware Ensemble via Task and Model Splitting
• When and How Does In-Distribution Label Help Out-of-Distribution Detection?

and 5 more in the interactive map.

Semi-supervised learning · 13 papers

• SAM as the Guide: Mastering Pseudo-Label Refinement in Semi-Supervised Referring Expression Segmentation (3 citations)
• InterLUDE: Interactions between Labeled and Unlabeled Data to Enhance Semi-Supervised Learning (2 citations)
• SimPro: A Simple Probabilistic Framework Towards Realistic Long-Tailed Semi-Supervised Learning (1 citation)
• Erasing the Bias: Fine-Tuning Foundation Models for Semi-Supervised Learning
• High-Dimensional Bayesian Optimization via Semi-Supervised Learning with Optimized Unlabeled Data Sampling
• OTMatch: Improving Semi-Supervised Learning with Optimal Transport
• Unveiling the Dynamics of Information Interplay in Supervised Learning
• Binary Decomposition: A Problem Transformation Perspective for Open-Set Semi-Supervised Learning
• Correlation-Induced Label Prior for Semi-Supervised Multi-Label Learning
• Fast Algorithms for Hypergraph PageRank with Applications to Semi-Supervised Learning
• Learning with Partial-Label and Unlabeled Data: A Uniform Treatment for Supervision Redundancy and Insufficiency
• SSL4Q: Semi-Supervised Learning of Quantum Data with Application to Quantum State Classification

and 1 more in the interactive map.

Zero-/few-shot and meta-learning · 13 papers

• Learning to Route Among Specialized Experts for Zero-Shot Generalization (1 citation)
• ELF: Encoding Speaker-Specific Latent Speech Feature for Speech Synthesis
• Learning Causal Domain-Invariant Temporal Dynamics for Few-Shot Action Recognition
• Let Go of Your Labels with Unsupervised Transfer
• MOKD: Cross-domain Finetuning for Few-shot Classification via Maximizing Optimized Kernel Dependence
• Sharpness-Aware Data Generation for Zero-shot Quantization
• Tilt your Head: Activating the Hidden Spatial-Invariance of Classifiers
• Unleashing the Power of Meta-tuning for Few-shot Generalization Through Sparse Interpolated Experts
• Enabling Few-Shot Learning with PID Control: A Layer Adaptive Optimizer
• Meta Evidential Transformer for Few-Shot Open-Set Recognition
• OT-CLIP: Understanding and Generalizing CLIP via Optimal Transport
• Prompting is a Double-Edged Sword: Improving Worst-Group Robustness of Foundation Models

and 1 more in the interactive map.

Knowledge distillation · 10 papers

• SelMatch: Effectively Scaling Up Dataset Distillation via Selection-Based Initialization and Partial Updates by Trajectory Matching
• What is Dataset Distillation Learning?
• DFD: Distilling the Feature Disparity Differently for Detectors
• Do Topological Characteristics Help in Knowledge Distillation?
• Enhancing Class-Imbalanced Learning with Pre-Trained Guidance through Class-Conditional Knowledge Distillation
• From Coarse to Fine: Enable Comprehensive Graph Self-supervised Learning with Multi-granular Semantic Ensemble
• Knowledge Distillation with Auxiliary Variable
• Large Scale Dataset Distillation with Domain Shift
• Revisit the Essence of Distilling Knowledge through Calibration
• USTAD: Unified Single-model Training Achieving Diverse Scores for Information Retrieval

Test-time adaptation · 7 papers

• Test-Time Model Adaptation with Only Forward Passes (3 citations)
• Evaluation of Test-Time Adaptation Under Computational Time Constraints (1 citation)
• The Entropy Enigma: Success and Failure of Entropy Minimization (1 citation)
• Multiply Robust Estimation for Local Distribution Shifts with Multiple Domains
• Domain-wise Data Acquisition to Improve Performance under Distribution Shift
• Stationary Latent Weight Inference for Unreliable Observations from Online Test-Time Adaptation
• TabLog: Test-Time Adaptation for Tabular Data Using Logic Rules

Noisy labels · 6 papers

• Don’t Label Twice: Quantity Beats Quality when Comparing Binary Classifiers on a Budget
• Learning with Complementary Labels Revisited: The Selected-Completely-at-Random Setting Is More Practical
• Pi-DUAL: Using privileged information to distinguish clean from noisy labels
• Does Label Smoothing Help Deep Partial Label Learning?
• Self-cognitive Denoising in the Presence of Multiple Noisy Label Sources
• Unbiased Multi-Label Learning from Crowdsourced Annotations