News
Contact me if interested in Postdoc or Student Research positions in scientific foundation models.
Research Interests
I am the AI Team Leader in the Information Sciences group at the Los Alamos National Laboratory. I develop AI and machine learning methods to help scientists analyze complex data. My research areas include foundation models for scientific discovery, trustworthy and reliable AI, AI safety and risk assessmet, probabilistic graphical models, unsupervised machine learning, interactive machine learning, transfer learning and lifelong machine learning. Applications include accelerating scientific simulations for discovery, AI safety, image analysis, and remote sensing.
- Project Leader / Thrust Leader:
- AI for Mission: Scientific Foundation Models (ArtIMis)
- AI Risk Assessment for WMD
- Project Leader / Principal Investigator (Past projects):
- Uncertainty quantification for robust machine learning (UQ4ML)
- Visual data analytics tools (VDAT)
- Image analysis using graphs (GoFigure)
- Applications of AI projects include spatio-temporal physics simulations, materials discovery, image analysis, ChemCam spectroscopy data analysis, and other scientific data analysis problems.
Publications
See longer publications list.
Data
- DeepPatent2 collection of 2.7M technical patent drawings with 132k object names and 22k viewpoints: dataset, paper
- DeepPatent collection of 350k technical patent drawings for image retrieval challenge: dataset, paper
Software
Open source code available for
- GoFigure collection for retrieval and analysis of technical images: GoFigure
- Visual Hash for matching copies of visually similar images: VisHash
- Visual Hash for matching copies of visually similar images: VisHash
- Bayesian discovery of multiple Bayesian networks: beandisco_multi
- Bayesian discovery of Bayesian networks with informative priors: beandisco_prior