Scientific agent swarms improve research by learning from each other
SwarmRSI: Recursive Self-Improvement through Evolving Scientific Agent Swarm
Artificial IntelligenceComputation and Language
Summary
Doing scientific research often means trying many ideas and experiments to find what works. The authors developed SwarmRSI, a system where groups of computer agents learn from their own and each other’s experiments to improve their research over time. These agents share feedback to fix ideas or their implementations, and combine useful parts from different agents to build better solutions. This makes it easier to improve the overall research performance and tackle new scientific problems more effectively. Experiments show that SwarmRSI can outperform previous methods and produce programs that biologists can reproduce for further study.
What this means in practice
- •For software developers: Develop research support tools that use agent swarms to automatically improve hypotheses and code based on experimental feedback.
- •For bioinformatics teams: Generate reproducible scientific programs to assist with complex biological research and follow-up analysis.
Authors
Qianru Meng, Xi Zhang, Konstantinos Papakostas, Congfeng Cao, Wenxiang Hu, Tianyi Lu, Xiao Zhang, Lun Zhou, Zaiqiao Meng, Zhaochun Ren, Xiaohui Yan, Yougang Lyu
Abstract
Recursive self-improvement (RSI) in scientific research aims to improve agents' research capabilities through their own experimental experience. Scientific agent swarms can combine evidence and implementations from different research directions to guide their agents' subsequent experiments. However, poor experimental results can leave unclear whether to revise a hypothesis or its implementation, while overall program scores can hide components useful to other agents. To address these challenges, we introduce SwarmRSI, a framework that evolves scientific agent swarms for recursive self-improvement. We devise Feedback-Guided Swarm Mutation to refine hypotheses and their implementations using experimental feedback from other agents. We also propose Complementary Swarm Recombination to reuse retained hypotheses and complementary program components across agents, allowing partial results to support new research. Recursive Swarm Evolution integrates both mechanisms into a feedback loop, using new experimental evidence to guide subsequent research. Experiments on OpenProblems, NatureBench and MLE-Bench Lite demonstrate SwarmRSI's state-of-the-art (SOTA) performance. Further evaluations show that accumulated research experience improves performance on held-out tasks. Biologists also reproduce its programs for follow-up biological analyses.