Scientific Discovery

Science advances when new instruments let us see what was previously out of reach. Our work on scientific discovery asks how machine learning can serve as such an instrument — and what it takes for its outputs to meet scientific standards of evidence.

Research topics

  • Scientific machine learning
  • Computational biology
  • Physics-informed learning
  • Chemistry and materials modeling
  • Simulation and surrogate models
  • Scientific reasoning
  • AI-assisted hypothesis generation
Fig. 07Schematic — not experimental data
ω₁ω₂
Fig. 07Schematic. Streamlines of a two-dimensional vector field with two sources of rotation — the kind of structure a learned surrogate must reproduce.

Why it matters

Why it matters

Many scientific problems are bottlenecked by computation: simulations that take weeks, search spaces too large to enumerate, or data too high-dimensional to inspect. Learned models can sometimes approximate these computations orders of magnitude faster, opening questions that were previously impractical to ask.

But scientific use imposes standards that ordinary benchmarks do not. Predictions must respect conservation laws and symmetries, come with honest error bars, and remain valid outside the regime they were trained on — or say clearly when they are not. Meeting those standards is a research problem in itself, and one that feeds back into the rest of machine learning.

Key research questions

Key research questions

  1. Q1

    How can known physical laws and symmetries be built into learned models without limiting what they can discover?

  2. Q2

    When can a learned surrogate safely replace a simulator, and how do we detect when it cannot?

  3. Q3

    What role can machine learning play in proposing hypotheses, as opposed to fitting data?

  4. Q4

    How should models represent biological and chemical systems that span many scales of space and time?

  5. Q5

    What does reproducibility mean for discoveries made with the assistance of learned models?

Primary subfields

Primary subfields

Physics-informed learning
Models constrained by differential equations, symmetries, and conservation laws.
Surrogate modeling
Learned approximations of expensive simulations with quantified error.
Computational biology
Sequence, structure, and systems-level modeling of living systems.
Chemistry and materials
Representations of molecules and materials for property prediction and design.
Hypothesis generation
Using models to propose, rank, and design experiments for scientific claims.

Representative methodologies

Representative methodologies

  • M1

    Equivariant architectures

    Networks whose outputs transform correctly under physical symmetries.

  • M2

    Simulation-based inference

    Inferring parameters of scientific models from observed data.

  • M3

    Active learning

    Choosing the most informative next experiment or simulation.

  • M4

    Benchmarking against established methods

    Comparing learned approaches with domain-standard numerical and experimental baselines.

Relationships to other domains

Relationships to other domains

Working on questions in Scientific Discovery?

We welcome conversations with researchers and organizations working on related questions — whether to compare notes, discuss a collaboration, or share work in progress.