NeurIPS 2019 accepted papers on Label-efficient and robust learning
95 of the 1,429 papers accepted at NeurIPS 2019 (6.6%) are on Label-efficient and robust learning, compared with 7.2% at NeurIPS 2018. They fall into 9 topics.
See NeurIPS 2019 in the interactive map
Submitting to NeurIPS? Checklist →
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 · 45 papers
- Cross-lingual Language Model Pretraining (1,596 citations)
- Transfusion: Understanding Transfer Learning for Medical Imaging (637 citations)
- This Looks Like That: Deep Learning for Interpretable Image Recognition (560 citations)
- On Mixup Training: Improved Calibration and Predictive Uncertainty for Deep Neural Networks (428 citations)
- A Benchmark for Interpretability Methods in Deep Neural Networks (389 citations)
- Learning Imbalanced Datasets with Label-Distribution-Aware Margin Loss (226 citations)
- Failing Loudly: An Empirical Study of Methods for Detecting Dataset Shift (188 citations)
- Glyce: Glyph-vectors for Chinese Character Representations (147 citations)
- Classification Accuracy Score for Conditional Generative Models (141 citations)
- Spatial-Aware Feature Aggregation for Image based Cross-View Geo-Localization (115 citations)
- A Meta-Analysis of Overfitting in Machine Learning (93 citations)
- AttentionXML: Label Tree-based Attention-Aware Deep Model for High-Performance Extreme Multi-Label Text Classification (92 citations)
and 33 more in the interactive map.
Zero-/few-shot and meta-learning · 11 papers
- Cross Attention Network for Few-shot Classification (139 citations)
- Multimodal Model-Agnostic Meta-Learning via Task-Aware Modulation (70 citations)
- Dual Adversarial Semantics-Consistent Network for Generalized Zero-Shot Learning (65 citations)
- Unsupervised Meta-Learning for Few-Shot Image Classification (65 citations)
- Transductive Zero-Shot Learning with Visual Structure Constraint (44 citations)
- Compositional generalization through meta sequence-to-sequence learning (43 citations)
- Zero-shot Learning via Simultaneous Generating and Learning (29 citations)
- Meta-Reinforced Synthetic Data for One-Shot Fine-Grained Visual Recognition (13 citations)
- Learning to Learn By Self-Critique (12 citations)
- Meta-Curvature (12 citations)
- Meta-Surrogate Benchmarking for Hyperparameter Optimization (12 citations)
Domain adaptation and generalization · 8 papers
- Domain Generalization via Model-Agnostic Learning of Semantic Features (426 citations)
- Learning Robust Global Representations by Penalizing Local Predictive Power (271 citations)
- Transferable Normalization: Towards Improving Transferability of Deep Neural Networks (132 citations)
- Category Anchor-Guided Unsupervised Domain Adaptation for Semantic Segmentation (93 citations)
- Multi-source Domain Adaptation for Semantic Segmentation (80 citations)
- Multi-marginal Wasserstein GAN (32 citations)
- Transfer Learning via Minimizing the Performance Gap Between Domains (28 citations)
- On the Value of Target Data in Transfer Learning (16 citations)
Self-supervised and contrastive · 8 papers
- Evaluating Protein Transfer Learning with TAPE (703 citations)
- Learning Representations by Maximizing Mutual Information Across Views (671 citations)
- Using Self-Supervised Learning Can Improve Model Robustness and Uncertainty (337 citations)
- Self-Supervised Generalisation with Meta Auxiliary Learning (93 citations)
- Putting An End to End-to-End: Gradient-Isolated Learning of Representations (79 citations)
- Self-supervised GAN: Analysis and Improvement with Multi-class Minimax Game (40 citations)
- Learning about an exponential amount of conditional distributions (12 citations)
- Information Competing Process for Learning Diversified Representations (8 citations)
Noisy labels · 6 papers
- Meta-Weight-Net: Learning an Explicit Mapping For Sample Weighting (398 citations)
- Are Anchor Points Really Indispensable in Label-Noise Learning? (188 citations)
- L_DMI: A Novel Information-theoretic Loss Function for Training Deep Nets Robust to Label Noise (120 citations)
- Correlated Uncertainty for Learning Dense Correspondences from Noisy Labels (20 citations)
- Combinatorial Inference against Label Noise (16 citations)
- On Single Source Robustness in Deep Fusion Models (15 citations)
Semi-supervised learning · 6 papers
- MixMatch: A Holistic Approach to Semi-Supervised Learning (595 citations)
- Learning to Self-Train for Semi-Supervised Few-Shot Classification (232 citations)
- Multi-label Co-regularization for Semi-supervised Facial Action Unit Recognition (31 citations)
- MarginGAN: Adversarial Training in Semi-Supervised Learning (27 citations)
- On Adversarial Mixup Resynthesis (23 citations)
- Semi-supervisedly Co-embedding Attributed Networks (8 citations)
Test-time adaptation · 5 papers
- Addressing Failure Prediction by Learning Model Confidence (169 citations)
- Fast and Flexible Multi-Task Classification using Conditional Neural Adaptive Processes (69 citations)
- Cold Case: The Lost MNIST Digits (63 citations)
- DTWNet: a Dynamic Time Warping Network (62 citations)
- Fixing the train-test resolution discrepancy (23 citations)
Knowledge distillation · 3 papers
- When does label smoothing help? (879 citations)
- Knowledge Extraction with No Observable Data (61 citations)
- Zero-shot Knowledge Transfer via Adversarial Belief Matching (33 citations)
