Research map / Learning theory and algorithms
Combinatorial and search optimization: research map
1,492 accepted papers on Combinatorial and search optimization in Learning theory and algorithms, from ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026), grouped into 4 clusters and 17 approaches. The busiest year so far is 2024.
Within Learning theory and algorithms, its share held steady from 15.6% in 2023–24 to 14.9% in 2025–26 (347 → 363 papers at ICML, NeurIPS, CVPR and AAAI, the venues with data for all four years).
Explore Combinatorial and search optimization in the interactive map
Working on something in this topic? Describe your idea in scime atlas to see which approach it falls under, the closest papers by meaning and how crowded the spot has become.
Approaches and key papers
objective · multi · solutions · 470 papers
Approaches in this cluster:
- Constrained neural network optimization (111 papers)
Embeds linear or orthogonal constraints into network layers and analyzes constrained training and routing. - Unsupervised neural combinatorial optimization (113 papers)
Learns solvers for combinatorial problems via unsupervised relaxation, divide-and-conquer and learned branching. - Decision-focused optimization (97 papers)
Integrates prediction with optimization, including data requirements, robustness and dynamic programming for budgeted search. - Multi-objective optimization (93 papers)
Finds Pareto solutions with learned partitions, preference-guided scalarization and evolutionary methods. - Globally optimal geometric optimization (56 papers)
Solves computer vision problems with certifiably optimal and distributed energy minimization methods.
Most cited and most cited since 2024:
- Multi-Task Learning as Multi-Objective Optimization (NeurIPS 2018 · 327 citations)
- ICE-BA: Incremental, Consistent and Efficient Bundle Adjustment for Visual-Inertial SLAM (CVPR 2018 · 153 citations)
- NN-Steiner: A Mixed Neural-Algorithmic Approach for the Rectilinear Steiner Minimum Tree Problem (AAAI 2024 · 11 citations)
- UDC: A Unified Neural Divide-and-Conquer Framework for Large-Scale Combinatorial Optimization Problems (NeurIPS 2024 · 10 citations)
logic · sat · boolean · 387 papers
Approaches in this cluster:
- Logic and answer set programming theory (150 papers)
Studies semantics, expressiveness and chase termination for logic-based knowledge representation. - Structure and complexity of constraint reasoning (92 papers)
Analyzes algorithms and hardness for answer set programming, symmetry breaking and graph counting. - SAT and model counting solvers (89 papers)
Improves satisfiability, model counting and pseudo-Boolean solving with learned heuristics and proof certification. - Differentiable and symbolic reasoning (56 papers)
Learns circuit-SAT, symbolic regression and neurosymbolic programs through differentiable approaches.
Most cited and most cited since 2024:
- Efficient Solvers for Minimal Problems by Syzygy-Based Reduction (CVPR 2017 · 142 citations)
- A Unified View of Piecewise Linear Neural Network Verification (NeurIPS 2018 · 133 citations)
- Adaptive Reactive Synthesis for LTL and LTLf Modulo Theories (AAAI 2024 · 10 citations)
- Global Lyapunov functions: a long-standing open problem in mathematics, with symbolic transformers (NeurIPS 2024 · 9 citations)
planning · search algorithm · tree search · 385 papers
Approaches in this cluster:
- Heuristic and local search algorithms (147 papers)
Develops bidirectional, anytime and local search with approximation guarantees. - Classical and HTN planning (145 papers)
Designs heuristics, compilations and symbolic search for numeric and hierarchical planning. - Monte Carlo tree search (55 papers)
Improves MCTS through state abstraction, speculation and best-arm identification. - Search for program synthesis (38 papers)
Guides program synthesis with learned and distribution-based search over candidate programs.
Most cited and most cited since 2024:
- Count-Based Exploration with Neural Density Models (ICML 2017 · 343 citations)
- Neural-Guided Deductive Search for Real-Time Program Synthesis from Examples (ICLR 2018 · 70 citations)
- MUVERA: Multi-Vector Retrieval via Fixed Dimensional Encoding (NeurIPS 2024 · 7 citations)
- Abstract Action Scheduling for Optimal Temporal Planning via OMT (AAAI 2024 · 6 citations)
integer · linear · mixed · 250 papers
Approaches in this cluster:
- Predict-then-optimize learning (89 papers)
Trains predictors with downstream linear or integer programs, including solver-free and multi-stage objectives. - Heuristics for integer programming (60 papers)
Solves large integer programs through neighborhood search, message passing and solver configuration learning. - Learning branch-and-cut for MILP (54 papers)
Learns branching, cutting and search policies for mixed-integer programs, with generalization guarantees. - Semidefinite relaxations for verification (47 papers)
Tightens neural network verification with SDP relaxations, bound propagation and branch-and-bound.
Most cited and most cited since 2024:
- DeepCoder: Learning to Write Programs (ICLR 2017 · 217 citations)
- Towards Fast Computation of Certified Robustness for ReLU Networks (ICML 2018 · 149 citations)
- An N-Point Linear Solver for Line and Motion Estimation with Event Cameras (CVPR 2024 · 9 citations)
- Scalable Neural Network Verification with Branch-and-bound Inferred Cutting Planes (NeurIPS 2024 · 8 citations)
Related topics in Learning theory and algorithms
- Matrix and tensor methods (1,074)
- Sample complexity (1,363)
- Fairness (516)
- Submodular and game algorithms (1,030)
- Prediction and decision losses (2,134)
- Kernels and regression theory (1,004)
- Clustering (725)
- Optimal transport (384)
