Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Energy landscapes for machine learning.

Andrew J Ballard1, Ritankar Das1, Stefano Martiniani1

  • 1University Chemical Laboratories, Lensfield Road, Cambridge CB2 1EW, UK. dw34@cam.ac.uk.

Physical Chemistry Chemical Physics : PCCP
|April 4, 2017
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Generative intelligence explores the chemical space of ten million catalysts.

Chemical science·2026
Same author

Accelerated Reaction Exploration across Scales: A Hybrid Operando and Modeling Study of Oxidation Kinetics in Monolayer Tungsten Disulfide.

Journal of the American Chemical Society·2026
Same author

Comparison of predictive approaches to the dynamics of activated catalytic processes.

Physical chemistry chemical physics : PCCP·2026
Same author

Comparison of Protein-Glycosaminoglycan Interactions in ff14sb/GLYCAM06j-1 and CHARMM36m Force Fields.

Journal of chemical information and modeling·2026
Same author

Energy Landscape Analysis of Membrane Proteins Using NMR-Based Hybrid Restraint Potentials.

Journal of chemical theory and computation·2026
Same author

Energy landscapes of the water hexamer and octamer for the MB-pol and TIP4P/2005 potentials.

The Journal of chemical physics·2026
Same journal

Lower bound of the capacitance of constant phase elements based on electrochemical impedance spectra.

Physical chemistry chemical physics : PCCP·2026
Same journal

Stability constants of lanthanide-nitrate complexes in aqueous solutions: a theoretical study.

Physical chemistry chemical physics : PCCP·2026
Same journal

Lead-free Cs<sub>3</sub>MnCl<sub>5</sub> and CsMnCl<sub>3</sub> crystals: rapid on-chip crystallization, phase transition and fluorescence sensing applications.

Physical chemistry chemical physics : PCCP·2026
Same journal

F-Interstitial passivation preserves host-like optoelectronic properties in <sup>229</sup>Th:YLF nuclear-clock platforms.

Physical chemistry chemical physics : PCCP·2026
Same journal

Structural trends of tryptophan dimer: hydrogen bonding <i>versus</i> π-stacking from an energy decomposition analysis perspective.

Physical chemistry chemical physics : PCCP·2026
Same journal

Achieving high thermoelectric performance in Sb<sub>2</sub>Se<sub>3</sub>-alloyed GeTe through synergistic optimization of electrical and thermal transport.

Physical chemistry chemical physics : PCCP·2026
See all related articles

Machine learning landscapes in physical sciences can be analyzed like molecular potential energy landscapes. This approach offers new insights into training, predictions, and emergent properties for interdisciplinary research.

Area of Science:

  • Physical sciences
  • Computational chemistry
  • Data science

Background:

  • Machine learning (ML) is increasingly used for non-linear fitting and prediction in physical sciences.
  • Fitting functions with multiple local minima create complex ML landscapes.

Purpose of the Study:

  • To apply molecular potential energy landscape analysis methods to ML landscapes.
  • To gain insights into ML training, solution spaces, and prediction characteristics.

Main Methods:

  • Analogy between ML landscapes and molecular potential energy landscapes.
  • Defining ML landscape properties analogous to molecular structure, thermodynamics, and kinetics.
  • Visualizing and analyzing the structure of ML landscapes.

Main Results:

Related Experiment Videos

  • ML landscapes can be visualized and analyzed using methods developed for molecular systems.
  • Analogies reveal emergent properties of ML models related to landscape structure.
  • New perspectives on understanding the behavior and predictions of ML models.

Conclusions:

  • Analyzing ML landscapes through a molecular analogy provides novel insights.
  • This interdisciplinary approach can enhance ML applications in physical sciences.
  • Suggests new research directions at the intersection of ML, physics, and chemistry.