MDA by Kevin Murphy: LLMs Meet Bayesian Inference for Scientific Discovery
sirbayes · x · 2026-08-12
A new paper by Kevin Murphy introduces the Model Discovery Agent (MDA), a framework combining LLMs with Bayesian experimental design. MDA uses an LLM as a proposer for candidate structures and standard Bayesian machinery like Sequential Monte Carlo (SMC) and Simulation-Based Inference (SBI) to discover latent mechanistic world models.
Key contributions include:
- MDA Framework: Couples LLMs with Bayesian methods for data-efficient discovery, extended to the M-open regime where the truth lies outside the current hypothesis class.
- New SOTA: Achieves state-of-the-art accuracy on existing physics and chemistry benchmarks with significantly fewer experiments.
- NeuronBench: Introduces a new partially-observed, stochastic electrophysiology benchmark.
The paper also explains how the learned summary avoids representational collapse by anchoring through a supervised objective, adapting freely to whatever channels the LLM proposes.
More from Research
- MLCommons launches MLPerf Endpoints, a benchmark that measures inference endpoints the way buyers actually purchase them — TheKanter · 2026-10-16
- DIY LEGO bench tests AI brick-building, GPT-6 Astra beats Opus — TheLegoless · 2026-10-16
- tszzl's theory: Tornado → Apache license → apaches → Southwest → Colorado — tszzl · 2026-10-16
- DIRECT: retrieval-based next-question suggestions for conversational agents — burkov · 2026-10-16
- A Qwen3 neuron tracks both Colorado and Tornado HTTP code; tszzl offers a theory — tszzl · 2026-10-16
- AI-Driven Timing-Aware Power Optimization Uses Learned Ranking With Baseline-Preserving Repair — Amit Sharma · 2026-10-16