CVEvolve: Autonomous Algorithm Discovery for Unstructured Scientific Data Processing

2026-05-11

AI Summary

This work addresses the challenge faced by domain scientists who often lack the expertise to develop specialized algorithms for processing unstructured scientific data characterized by high noise levels, large dynamic ranges, sparse annotations, or ambiguous definitions. To bridge this gap, we propose the first zero-code, autonomous algorithm discovery system tailored for scientific data analysis. Built upon a large language model–driven agent architecture, the system integrates lineage-aware random sampling with multi-round iterative search to automatically generate, execute, evaluate, and visualize algorithms while managing historical trajectories. It effectively balances exploration and exploitation during the search process. Evaluated on tasks including X-ray fluorescence microscopy image registration, Bragg peak detection, and high-energy diffraction microscopy image segmentation, the system consistently produces algorithms that significantly outperform baseline methods and demonstrate superior generalization on held-out test sets.

Abstract

Scientific data processing often requires task-specific algorithms or AI models, creating a barrier for domain scientists who need to analyze their data but may not have extensive computing or image-processing expertise. This barrier is especially pronounced when data are noisy, have a high dynamic range, are sparsely labeled, or are only loosely specified. We introduce CVEvolve, an autonomous agentic harness with a zero-code interface for scientific data-processing algorithm discovery. CVEvolve combines a multi-round search strategy with tools for code execution, evaluation implementation, history management, holdout testing, and optional inspection of scientific data and visual outputs. The search alternates between discovery and improvement actions, and uses lineage-aware stochastic candidate sampling to balance exploration and exploitation. We demonstrate CVEvolve on x-ray fluorescence microscopy image registration, Bragg peak detection, and high-energy diffraction microscopy image segmentation. Across these tasks, CVEvolve discovers algorithms that improve over baseline methods, while holdout test tracking helps identify candidates that generalize better than later over-optimized alternatives. These results show that zero-code, autonomous LLM-powered algorithm development can help domain scientists turn unstructured scientific image data into practical algorithms and downstream scientific discoveries.

Problem

• scientific data processing
• unstructured data
• algorithm discovery
• domain scientists
• zero-code interface

Innovation

• autonomous algorithm discovery
• zero-code interface
• lineage-aware sampling
• scientific image processing
• LLM-powered agent