Problem
Research questions and friction points this paper is trying to address.
scientific idea generation
information cocoons
idea diversity
static retrieval-generation paradigm
autonomous research
FlowPIE: Test-Time Scientific Idea Evolution with Flow-Guided Literature Exploration
🤖 AI Summary
Existing methods for scientific idea generation are constrained by static retrieval-generation paradigms, often yielding homogeneous and insufficiently diverse outputs. This work reframes literature exploration and idea generation as a co-evolutionary process and introduces a novel framework that integrates flow-guided Monte Carlo Tree Search (MCTS) with an island-model evolutionary mechanism. Inspired by Generative Flow Networks (GFlowNets), the approach employs flow-guided MCTS to construct a high-quality initial population and enables dynamic evolution at test time through genetic operators—selection, crossover, and mutation—to facilitate cross-domain knowledge integration. Notably, it is the first method to support dynamic reward scaling during inference, significantly enhancing the novelty, feasibility, and diversity of generated ideas and outperforming current large language model– and agent-based approaches.
📝 Abstract
Scientific idea generation (SIG) is critical to AI-driven autonomous research, yet existing approaches are often constrained by a static retrieval-then-generation paradigm, leading to homogeneous and insufficiently divergent ideas. In this work, we propose FlowPIE, a tightly coupled retrieval-generation framework that treats literature exploration and idea generation as a co-evolving process. FlowPIE expands literature trajectories via a flow-guided Monte Carlo Tree Search (MCTS) inspired by GFlowNets, using the quality of current ideas assessed by an LLM-based generative reward model (GRM) as a supervised signal to guide adaptive retrieval and construct a diverse, high-quality initial population. Based on this population, FlowPIE models idea generation as a test-time idea evolution process, applying selection, crossover, and mutation with the isolation island paradigm and GRM-based fitness computation to incorporate cross-domain knowledge. It effectively mitigates the information cocoons arising from over-reliance on parametric knowledge and static literature. Extensive evaluations demonstrate that FlowPIE consistently produces ideas with higher novelty, feasibility and diversity compared to strong LLM-based and agent-based frameworks, while enabling reward scaling during test time.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Flow-guided MCTS
Generative Reward Model
Test-Time Idea Evolution
Scientific Idea Generation
Island Paradigm
🔎 Similar Papers
No similar papers found.
