Heterogeneous agentic coevolution improves specialized ai models
Heterogeneous Agentic Coevolution
Artificial Intelligence
Summary
Many AI systems that focus on specific tasks rely on specialized models that need humans to help improve them, which can slow progress. The authors present a method called Heterogeneous Agentic CoEvolution (HACE) where two different AI models learn from each other to get better together. One model specializes in the task, while the other explores ways to improve it, with both adapting as each gets stronger. They tested this approach in areas like recommending products, predicting protein functions, and discovering new materials, finding improvements over previous methods.
What this means in practice
- •For recommendation system engineers: Improve recommendation algorithms by enabling adaptive learning between specialized models and exploration agents for better suggestions.
- •For materials discovery teams: Enhance crystal stability predictions by evolving specialized models jointly with assisting agents to speed material innovation.
Authors
Wenqian Ye, Ziwei Guan, Zi Yang, Shivani Modi, Cici Cong Wu, Yunlong He, Yuxin Chen, Zhengyu Zhang, Liang Luo, Venkatesh Ranganathan, Jianhui Sun, Yantao Yao, Wenlin Chen, Buyun Zhang, Ellie Dingqiao Wen, Henry Kautz, Aidong Zhang
Abstract
Recent advances in self-improving LLM agents have demonstrated the potential of AI systems that drive their own intelligence. However, many domain-specific applications rely on specialized foundation models that possess strong domain expertise but remain dependent on human-designed improvements. Extending recursive self-improvement to these models presents a key challenge: as a specialist improves, the strategies used to advance it must also adapt. To address this challenge, we propose Heterogeneous Agentic CoEvolution (HACE), a framework that jointly evolves specialized foundation models and LLM agents through reciprocal learning. HACE improves the specialist through supervised learning while training the agent with on-policy reinforced self-training, guided by verified performance gains on the current specialist. The agent's exploration, in turn, informs subsequent specialist training, enabling both models to adapt to each other's progress. To sustain this reciprocal learning, HACE further employs evolutionary search to optimize their interaction harness, adapting how the models exchange information and select training experiences as their capabilities evolve. We evaluate HACE across three challenging domains: recommendation systems, protein function prediction in biology, and crystal stability prediction in material discovery. We show that HACE outperforms the evaluated self-improving methods on the primary task metrics across all three domains. These results shed light on heterogeneous coevolution as a universal approach to improving specialized foundation models through reciprocal interaction with a learning agent.