Agentic Automated Algorithm Search for Quantum-computing-integrated Machine Learning

Track: Track 1: Technical Foundations for a Post-AGI World

Keywords: Agentic framework, Reasoning, LLM, Machine Learning, Quantum Integrated Machine Learning

Abstract: By leveraging pretrained LLM, we present a LLM-based multi-agent framework that automatically searches in program space to synthesize/find quantum-computing-integrated machine learning algorithms from classical origins. Our agentic system operates as an orchestrated set of agents with capabilities of tool use, integration of internet knowledge, and planning. The searching workflow then utilizes Monte Carlo Tree Search for guided exploration, a nested dual-loop procedure for robust code generation and debugging, as well as a Webscraper for up-to-date knowledge injection. Set a algorithmic seed as a classical multi-layer perceptron, the system generates quantum-hybrid and quantum-inspired models by iteratively proposing, validating, and refining code. With the empirical evidence, the system can find novel algorithms with high performance. By coupling LLM reasoning and coding with principled search and verification, our approach offers a practical path toward accelerated design of algorithms for quantum-computing-integrated machine learning.

Anonymization: This submission has been anonymized for double-blind review via the removal of identifying information such as names, affiliations, and identifying URLs.

Presenter: ~FuTe_Wong1

Format: Maybe: the presenting author will attend in person, contingent on other factors that still need to be determined (e.g., visa, funding).

Funding: Yes, the presenting author of this submission falls under ICLR’s funding aims, and funding would significantly impact their ability to attend the workshop in person.

Submission Number: 32