NeurIPS 2020 accepted papers on Learning theory and algorithms
363 of the 1,901 papers accepted at NeurIPS 2020 (19.1%) are on Learning theory and algorithms, compared with 19.5% at NeurIPS 2019. They fall into 9 topics.
See NeurIPS 2020 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
Prediction and decision losses · 75 papers
- Neural Networks Fail to Learn Periodic Functions and How to Fix It (55 citations)
- Large-Scale Methods for Distributionally Robust Optimization (49 citations)
- Coupling-based Invertible Neural Networks Are Universal Diffeomorphism Approximators (44 citations)
- Curriculum Learning by Dynamic Instance Hardness (43 citations)
- Model Interpretability through the lens of Computational Complexity (38 citations)
- WoodFisher: Efficient Second-Order Approximation for Neural Network Compression (36 citations)
- Beyond Individualized Recourse: Interpretable and Interactive Summaries of Actionable Recourses (32 citations)
- Learning with Differentiable Pertubed Optimizers (32 citations)
- Deep Smoothing of the Implied Volatility Surface (31 citations)
- A meta-learning approach to (re)discover plasticity rules that carve a desired function into a neural network (27 citations)
- A mathematical model for automatic differentiation in machine learning (24 citations)
- NeuMiss networks: differentiable programming for supervised learning with missing values. (23 citations)
and 63 more in the interactive map.
Sample complexity · 70 papers
- What Neural Networks Memorize and Why: Discovering the Long Tail via Influence Estimation (115 citations)
- On the Theory of Transfer Learning: The Importance of Task Diversity (65 citations)
- Sample Complexity of Asynchronous Q-Learning: Sharper Analysis and Variance Reduction (59 citations)
- Universally Quantized Neural Compression (47 citations)
- Exactly Computing the Local Lipschitz Constant of ReLU Networks (32 citations)
- Rational neural networks (24 citations)
- Task-Robust Model-Agnostic Meta-Learning (24 citations)
- On the Modularity of Hypernetworks (22 citations)
- Near-Optimal SQ Lower Bounds for Agnostically Learning Halfspaces and ReLUs under Gaussian Marginals (19 citations)
- Statistical and Topological Properties of Sliced Probability Divergences (18 citations)
- Neural Networks Learning and Memorization with (almost) no Over-Parameterization (17 citations)
- Towards Understanding Hierarchical Learning: Benefits of Neural Representations (16 citations)
and 58 more in the interactive map.
Kernels and regression theory · 51 papers
- Sharpened Generalization Bounds based on Conditional Mutual Information and an Application to Noisy, Iterative Algorithms (48 citations)
- A random matrix analysis of random Fourier features: beyond the Gaussian kernel, a precise phase transition, and the corresponding double descent (40 citations)
- Kernel Methods Through the Roof: Handling Billions of Points Efficiently (40 citations)
- Exact expressions for double descent and implicit regularization via surrogate random design (38 citations)
- Generalization error in high-dimensional perceptrons: Approaching Bayes error with convex optimization (30 citations)
- On the Similarity between the Laplace and Neural Tangent Kernels (30 citations)
- Why Do Deep Residual Networks Generalize Better than Deep Feedforward Networks? --- A Neural Tangent Kernel Perspective (29 citations)
- Spectra of the Conjugate Kernel and Neural Tangent Kernel for linear-width neural networks (27 citations)
- Triple descent and the two kinds of overfitting: where & why do they appear? (27 citations)
- Kernel Alignment Risk Estimator: Risk Prediction from Training Data (26 citations)
- A Measure-Theoretic Approach to Kernel Conditional Mean Embeddings (25 citations)
- Autoencoders that don't overfit towards the Identity (23 citations)
and 39 more in the interactive map.
Matrix and tensor methods · 47 papers
- Implicit Regularization in Deep Learning May Not Be Explainable by Norms (61 citations)
- Phase retrieval in high dimensions: Statistical and computational phase transitions (21 citations)
- Practical Low-Rank Communication Compression in Decentralized Deep Learning (20 citations)
- Node Embeddings and Exact Low-Rank Representations of Complex Networks (17 citations)
- Adversarial Crowdsourcing Through Robust Rank-One Matrix Completion (15 citations)
- Matrix Completion with Quantified Uncertainty through Low Rank Gaussian Copula (15 citations)
- All-or-nothing statistical and computational phase transitions in sparse spiked matrix estimation (14 citations)
- Robust Sub-Gaussian Principal Component Analysis and Width-Independent Schatten Packing (12 citations)
- The All-or-Nothing Phenomenon in Sparse Tensor PCA (12 citations)
- Outlier Robust Mean Estimation with Subgaussian Rates via Stability (11 citations)
- Coded Sequential Matrix Multiplication For Straggler Mitigation (9 citations)
- Estimating Rank-One Spikes from Heavy-Tailed Noise via Self-Avoiding Walks (8 citations)
and 35 more in the interactive map.
Combinatorial and search optimization · 37 papers
- AI Feynman 2.0: Pareto-optimal symbolic regression exploiting graph modularity (85 citations)
- On the Role of Sparsity and DAG Constraints for Learning Linear DAGs (58 citations)
- On Warm-Starting Neural Network Training (53 citations)
- Interior Point Solving for LP-based prediction+optimisation (42 citations)
- The Convex Relaxation Barrier, Revisited: Tightened Single-Neuron Relaxations for Neural Network Verification (42 citations)
- Mitigating Manipulation in Peer Review via Randomized Reviewer Assignments (33 citations)
- Certified Monotonic Neural Networks (29 citations)
- DAGs with No Fears: A Closer Look at Continuous Optimization for Learning Bayesian Networks (28 citations)
- Benchmarking Deep Inverse Models over time, and the Neural-Adjoint method (20 citations)
- Belief Propagation Neural Networks (19 citations)
- A Scalable MIP-based Method for Learning Optimal Multivariate Decision Trees (17 citations)
- Semialgebraic Optimization for Lipschitz Constants of ReLU Networks (17 citations)
and 25 more in the interactive map.
Submodular and game algorithms · 29 papers
- Chaos, Extremism and Optimism: Volume Analysis of Learning in Games (21 citations)
- Decisions, Counterfactual Explanations and Strategic Behavior (14 citations)
- Monotone operator equilibrium networks (13 citations)
- Near-Optimal Reinforcement Learning with Self-Play (13 citations)
- No-Regret Learning and Mixed Nash Equilibria: They Do Not Mix (13 citations)
- The Smoothed Possibility of Social Choice (13 citations)
- Deterministic Approximation for Submodular Maximization over a Matroid in Nearly Linear Time (11 citations)
- Optimally Deceiving a Learning Leader in Stackelberg Games (11 citations)
- Pipeline PSRO: A Scalable Approach for Finding Approximate Nash Equilibria in Large Games (10 citations)
- Real World Games Look Like Spinning Tops (10 citations)
- Hitting the High Notes: Subset Selection for Maximizing Expected Order Statistics (8 citations)
- Polynomial-Time Computation of Optimal Correlated Equilibria in Two-Player Extensive-Form Games with Public Chance Moves and Beyond (8 citations)
and 17 more in the interactive map.
Optimal transport · 20 papers
- Distribution Matching for Crowd Counting (169 citations)
- COT-GAN: Generating Sequential Data via Causal Optimal Transport (41 citations)
- Entropic Optimal Transport between Unbalanced Gaussian Measures has a Closed Form (41 citations)
- A Universal Approximation Theorem of Deep Neural Networks for Expressing Probability Distributions (33 citations)
- Partial Optimal Tranport with applications on Positive-Unlabeled Learning (30 citations)
- Differentiable Top-k with Optimal Transport (26 citations)
- Asymptotic Guarantees for Generative Modeling Based on the Smooth Wasserstein Distance (14 citations)
- Linear Time Sinkhorn Divergences using Positive Features (13 citations)
- Statistical Optimal Transport posed as Learning Kernel Embedding (9 citations)
- Quantifying the Empirical Wasserstein Distance to a Set of Measures: Beating the Curse of Dimensionality (8 citations)
- Sufficient dimension reduction for classification using principal optimal transport direction (8 citations)
- Model Fusion via Optimal Transport (7 citations)
and 8 more in the interactive map.
Fairness · 19 papers
- Robust Optimization for Fairness with Noisy Protected Groups (40 citations)
- Learning Certified Individually Fair Representations (28 citations)
- Achieving Equalized Odds by Resampling Sensitive Attributes (22 citations)
- Ensuring Fairness Beyond the Training Data (19 citations)
- A Fair Classifier Using Kernel Density Estimation (16 citations)
- How do fair decisions fare in long-term qualification? (15 citations)
- Neutralizing Self-Selection Bias in Sampling for Sortition (8 citations)
- Metric-Free Individual Fairness in Online Learning (6 citations)
- Trade-offs and Guarantees of Adversarial Representation Learning for Information Obfuscation (6 citations)
- Approximate Heavily-Constrained Learning with Lagrange Multiplier Models (5 citations)
- Exploiting MMD and Sinkhorn Divergences for Fair and Transferable Representation Learning
- Fair Multiple Decision Making Through Soft Interventions
and 7 more in the interactive map.
Clustering · 15 papers
- Efficient Clustering Based On A Unified View Of $K$-means And Ratio-cut (14 citations)
- Sliding Window Algorithms for k-Clustering Problems (12 citations)
- Discovering conflicting groups in signed networks (9 citations)
- Improved Guarantees for k-means++ and k-means++ Parallel (8 citations)
- Simple and Scalable Sparse k-means Clustering via Feature Ranking (8 citations)
- Efficient Clustering for Stretched Mixtures: Landscape and Optimality (4 citations)
- Exact Recovery of Mangled Clusters with Same-Cluster Queries (3 citations)
- KFC: A Scalable Approximation Algorithm for $k$−center Fair Clustering (3 citations)
- Higher-Order Spectral Clustering of Directed Graphs (2 citations)
- Probabilistic Fair Clustering (2 citations)
- Ratio Trace Formulation of Wasserstein Discriminant Analysis (1 citation)
- Near-Optimal Comparison Based Clustering
and 3 more in the interactive map.
