Researchers have introduced EPOCH, a novel framework for automated heuristic design (AHD) that leverages Large Language Models (LLMs) to create heuristics for combinatorial optimization problems. Unlike previous methods that either freeze the LLM or fine-tune it on a single fitness score, EPOCH optimizes the evolutionary search to maximize the coverage of training instances by the generated heuristics. The framework also incorporates a mechanism for the LLM generator to learn from its own search process through experiential preference optimization, using natural-language feedback from a teacher LLM to refine its heuristic generation capabilities. This approach results in complementary heuristic ensembles that outperform existing baselines and generalize across different problem scales. AI
IMPACT This research could lead to more efficient and adaptable heuristic design for complex optimization problems, potentially impacting fields reliant on such solutions.
RANK_REASON The cluster describes a research paper detailing a new method for automated heuristic design using LLMs. [lever_c_demoted from research: ic=1 ai=1.0]
- alphaXiv
- arXiv
- CatalyzeX
- Combinatorial Optimization Problems
- DagsHub
- Direct Preference Optimization: Your Language Model is Secretly a Reward Model
- EPOCH
- Gotit.pub
- Hugging Face
- Reflective Self-Distillation
- ScienceCast
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
