Artificial intelligence and machine learning
The Top Open Problems editorial order restricted to Computer Science, followed by cross-listed economics entries. Existing relative order is retained, with provisional placements for new game-learning questions; most entries have no individual difficulty assessment. About rankings
Browse by subject
Search this collection by problem ID, title, classification, subject, provider, or model.
Search results
No matching problems. Try another search or clear a filter.
54 records · Search and filter
- CS-50 Polynomial-Time Distribution-Free PAC Learning of DNF Formulas
- CS-66 Sample Compression Conjecture (Linear Size in VC Dimension)
- CS-179 10 Lectures and 42 Open Problems — Boolean classification and annulus conjecture
- CS-180 10 Lectures and 42 Open Problems — Tightness of the Multireference Alignment SDP
- CS-189 10 Lectures and 42 Open Problems — Consistency and sample complexity of Multireference Alignment
- CS-333 Accurate Answers to Adaptively Chosen Statistical Queries
- CS-348 Global Optimization of Overparameterized Shallow Networks
- CS-349 Extrapolating Clusterings from Semidefinite Subsamples
- CS-394 Optimal Neural-Network Depth for Sparse Recovery
- CS-410 Universal Approximation by One-Lipschitz Spline Networks
- CS-411 Global Margin Maximization beyond the Neural Tangent Regime
- CS-412 Cancellation-Sensitive Norms for Two-Layer Networks
- CS-414 Geometric Criteria for Benign Overfitting
- CS-415 Benign, Tempered, and Catastrophic Overfitting Regimes
- CS-426 Semi-Supervised Sparse Gaussian Classification
- CS-447 Cross-Entropy Dynamics Beyond Hadamard Initialization
- CS-448 Hidden-Layer Depth Complexity of Piecewise-Linear Functions
- CS-656 Learning Capacity of Totally Geodesic Map Families
- CS-678 Machine-Learning Algorithms Results Report
- CS-681 Robust Spectral Tensor Methods for Latent-Variable Learning
- CS-710 Provable Advantages of Convolutional Arithmetic Circuits
- CS-726 Robustness-Driven Overparameterization in Two-Layer Neural Networks
- CS-728 Comparative Advantages of Physics-Based and Machine-Learned Models
- CS-729 High-Dimensional Regression and PDE Learning Without Dimensional Curse
- CS-740 Computable Neural-Network Approximation from Noisy Samples
- CS-741 Computable Ground Truths for Learning Problems
- CS-742 Stochasticity and Sample Complexity in Neural Networks
- CS-751 Neural Network Robustness under Adversarial and Distribution Shifts
- CS-757 Mathematical Conjecture Generation with Machine Intelligence
- CS-759 Training-Loss Implications for Shallow Neural Networks
- CS-760 Convergence to Implicit Geometry under Class Imbalance
- CS-761 Neural Approximation of Strongly Convex Drifts with Linear Error
- CS-768 Expressivity and Training Theory for Deep Material Networks
- CS-771 Learning Data Manifolds for Augmentation and Nonlinear Components
- CS-775 Autoencoder Optimization Perspectives
- CS-776 Optimal Noiseless Rates for Max-Average-Margin Classifiers
- CS-779 Neural Networks for High-Dimensional Optimal Feedback Control
- CS-789 Generalization Guarantees for Self-Supervised Representations
- CS-790 Adam Optimization Performance
- CS-791 Definition of a Causal Representation
- CS-792 Criteria for a Good Representation
- CS-807 Efficient Optimization of Neural-Network Weights
- CS-811 Formal Verification of AI-Generated Mathematical Proofs
- CS-813 Agentic and Neuro-Symbolic AI for Mathematical Research
- CS-815 Causal-Attention Retrieval Motivation
- CS-817 Complexity and Generalization in Overparameterized Learning
- CS-818 Unified Mathematical Theory of Modern Learning
- CS-819 Curricula for Algorithmic Reasoning in Sequence Models
- CS-820 Inference and Training Effects of the Softmax Bottleneck
- CS-821 Extremal Top-$m$ Representability of Output Projection Matrices