NIPS-20 待读的Paper

2020.10.6

NIPS-2020 1900 accepted papers

(一般在 author notification 后过几天放出来）

粗略搜索了一些相关的论文

Interesting paper

• Self-Distillation as Instance-Specific Label Smoothing
• Provably Consistent Partial-Label Learning
• Learning from Label Proportions: A Mutual Contamination Framework
• Rethinking Importance Weighting for Deep Learning under Distribution Shift

Noisy labels

• Parts-dependent Label Noise: Towards Instance-dependent Label Noise
• Dual T: Reducing Estimation Error for Transition Matrix in Label-noise Learning
• Identifying Mislabeled Data using the Area Under the Margin Ranking
• Coresets for Robust Training of Deep Neural Networks against Noisy Labels
• Early-Learning Regularization Prevents Memorization of Noisy Labels
• A Topological Filter for Learning with Label Noise
• What Do Neural Networks Learn When Trained With Random Labels?

Class-imbalance Learning

• MESA: Effective Ensemble Imbalanced Learning with MEta-SAmpler

• Posterior Re-calibration for Imbalanced Datasets

• Generative Modeling of Factorized Representations in Class-Imbalanced Data

• Rethinking the Value of Labels for Improving Class-Imbalanced Learning

• Devil in the Momentum: Long-Tailed Classification by Removing Momentum Causal Effect

• Balanced Meta-Softmax for Long-Tailed Visual Recognition

• What Neural Networks Memorize and Why: Discovering the Long Tail via Influence Estimation

• Fast Unbalanced Optimal Transport on Tree

• Distribution Aligning Refinery of Pseudo-label for Imbalanced Semi-supervised Learning (glz 12-组会)

PU Learning

• Learning from Positive and Unlabeled Data with Arbitrary Positive Shift
• A Variational Approach for Learning from Positive and Unlabeled Data
• Semi-Supervised Partial Label Learning via Confidence-Rated Margin Maximization
• Temporal Positive-unlabeled Learning for Biomedical Hypothesis Generation via Risk Estimation
• Partial Optimal Transport with applications on Positive-Unlabeled Learning

Domain Adaptation:

• Domain Adaptation with Conditional Distribution Matching and Generalized Label Shift

Model calibration

• Improving model calibration with accuracy versus uncertainty optimization

点云，3D 重建

• CaSPR: Learning Canonical Spatiotemporal Point Cloud Representations
• Group Contextual Encoding for 3D Point Clouds
• PIE-NET: Parametric Inference of Point Cloud Edges
• Rotation-Invariant Local-to-Global Representation Learning for 3D Point Cloud
• Self-Supervised Few-Shot Learning on Point Clouds