ICML2022有意思的文章

ICML2022放榜了，很久没写博客了，今天摸鱼把ICML2022的接收列表翻了一遍，挑了一些自己感兴趣的下次读, 先放论文列表，有兴趣的自取。

Oral

• Head2Toe: Utilizing Intermediate Representations for Better Transfer Learning
• Bounding Training Data Reconstruction in Private (Deep) Learning
• Measuring Representational Robustness of Neural Networks Through Shared Invariances
• ModLaNets: Learning Generalisable Dynamics via Modularity and Physical Inductive Bias
• Hierarchical Shrinkage: Improving the accuracy and interpretability of tree-based models

Spotlights

• Does the Data Induce Capacity Control in Deep Learning?
• Accelerating Shapley Explanation via Contributive Cooperator Selection
• Prototype Based Classification from Hierarchy to Fairness
• Measuring the Effect of Training Data on Deep Learning Predictions via Randomized Experiments
• Fair Representation Learning through Implicit Path Alignment
• Feature selection using e-values
• Mitigating Neural Network Overconfidence with Logit Normalization
• Active Multi-Task Representation Learning
• Dataset Condensation with Contrastive Signals
• Extracting Latent State Representations with Linear Dynamics from Rich Observations
• How to Fill the Optimal Set? Population Gradient Descent with Harmless Diversity
• Fair and Fast k-Center Clustering for Data Summarization
• Channel Importance Matters in Few-Shot Image Classification
• Label-Free Explainability for Unsupervised Models
• A psychological theory of explainability
• From data to functa: Your data point is a function and you should treat it like one
• Understanding Robust Overfitting of Adversarial Training and Beyond
• Learning Stable Classifiers by Transferring Unstable Features
• Interpretable Neural Networks with Frank-Wolfe: Sparse Relevance Maps and Relevance Orderings
• XAI for Transformers: Better Explanations through Conservative Propagation
• Role-based Multiplex Network Embedding
• Meaningfully debugging model mistakes using conceptual counterfactual explanations
• Forgetting-free Continual Learning with Winning Subnetworks
• Wide Neural Networks Forget Less Catastrophically
• Measuring dissimilarity with diffeomorphism invariance
• Efficient Learning of CNNs using Patch Based Features
• Multi-scale Feature Learning Dynamics: Insights for Double Descent

Accepted papers

• Achieving Fairness at No Utility Cost via Data Reweighing
• Confidence Score for Source-Free Unsupervised Domain Adaptation
• Probabilistic Bilevel Coreset Selection
• Transfer and Marginalize: Explaining Away Label Noise with Privileged Information
• On the Effects of Artificial Data Modification
• Provable Domain Generalization via Invariant-Feature Subspace Recovery
• More Than a Toy: Random Matrix Models Predict How Real-World Neural Representations Generalize
• Information-Intensive Dataset Condensation
• Datamodels: Understanding Predictions with Data and Data with Predictions
• Benefits of Deep and Wide Convolutional Residual Networks: Function Approximation under Smoothness Constraint
• What Can Linear Interpolation of Neural Network Loss Landscapes Tell Us?
• Understanding Instance-Level Impact of Fairness Constraints
• Disentangling Disease-related Representation from Obscure for Disease Prediction
• A new similarity measure for covariate shift with applications to nonparametric regression
• Representation Topology Divergence: A Method for Comparing Neural Network Representations