AlgoSkill: Learning to Design Algorithms by Scheduling Human-Like Skills
2026-06-29 • Artificial Intelligence
Artificial Intelligence
AI summaryⓘ
The authors created AlgoSkill, a method that breaks down designing algorithms from natural language into smaller, clear steps called skills. Instead of just generating code at once, AlgoSkill carefully picks and tries out these skills to build better algorithms, checking correctness and improving complexity along the way. Their tests show this approach works better than simpler code generation methods. They found that using specific skill types, verifying solutions, and smartly searching through options all help make the process more reliable.
Algorithm designLarge language modelsMonte Carlo Tree SearchVerificationConstraint analysisData structuresSelf-refinementComplexity analysisAbstractionSkill scheduling
Authors
Xinyuan Song, Zekun Cai, Liang Zhao
Abstract
Designing an algorithm from a natural-language problem statement requires identifying the problem structure, reading constraints, choosing a suitable paradigm, checking correctness, and refining complexity. Existing large language model (LLM) methods often rely on direct generation or generic self-refinement, leaving these steps implicit. We propose AlgoSkill, which models algorithm design as sequential decision-making over a typed library of algorithmic skills, including abstraction, constraint analysis, state design, data-structure selection, proof checking, counterexample construction, and complexity refinement. A learned scheduler proposes skills from the current design state, while a Monte Carlo Tree Search (MCTS) controller explores skill sequences using verification feedback from compilation, testing, stress testing, and complexity analysis. Experiments on competitive programming and combinatorial optimization benchmarks show that AlgoSkill improves over direct LLM generation, chain-of-thought prompting, self-refinement, and MCTS without typed skills. Ablations show that typed skills, verification-based repair, and search-based scheduling each contribute to performance. These results support treating automatic algorithm design as verification-guided skill scheduling rather than one-shot code generation.