Science in the Age of Foundation Models
Talk, Siemens PhysicalAI Research Seminar, Siemens
Talks and presentations
Science in the Age of Foundation Models
Talk, MEMS Department Seminar Series, University of Pittsburgh
I spoke at the MEMS Department Seminar Series at the University of Pittsburgh. Please find my slides here.
Science in the Age of Foundation Models
Talk, Deep Learning for Science School, University of California, Berkeley
I will be co-organizing and speaking at the Deep Learning for Science School on our work on physics foundation models and constraints at the University of California, Berkeley.
Understanding the Bitter Lesson in Time Series Foundation Models
Talk, Computational Math in Industry and Beyond Seminar Series (CME 500), Stanford, CA
I gave an invited talk at the Computational Math in Industry and Beyond Seminar Series (CME 500) at the Institute for Computational and Mathematical Engineering (ICME) at Stanford University on our work on the development of time series foundation models, including Chronos and Chronos-Bolt. In particular, I discussed how subtle differences in their design choices can lead to inductive biases and performance differences on various time series datasets including chaotic systems. Please find my slides here.
Commencement Speaker
Talk, St. Joseph Notre Dame High School, Alameda, CA
I was the commencement speaker at the graduation ceremony at my past high school, St. Joseph Notre Dame, where I gave the salutatorian address when I graduated in 2008. I spoke on how I built a career in math and led research initiatives in AI for Science, and gave advice for their futures. Please find my speech here.
Science in the Age of Foundation Models
Talk, University of California, Santa Cruz, Santa Cruz, CA
I spoke at the University of California, Santa Cruz Applied Math Seminar on our work on foundation models for scientific applications including weather forecasting, automotive aerodynamics and chaotic systems. Please find my slides here.
LLMs for Time Series Forecasting
Talk, BERTS NeurIPS Workshop, San Diego, CA
I spoke at the BERTS NeurIPS workshop on our recent advances in time series foundation models with our Chronos models and their applications to chaotic systems forecasting. Please find my slides here.
Understanding the Bitter Lesson In Time Series Foundation Models
Talk, Amazon Supply Chain Optimization Technologies (SCOT) Deep Learning Science Seminar, Palo Alto, CA
I spoke at the Amazon SCOT Deep Learning Science Seminar on our work “Understanding the Implicit Biases of Design Choices for Time Series Foundation Models”.
Science in the Age of Foundation Models
Talk, Nxt AI, San Francisco, CA
I spoke at the Nxt AI conference on our recent advances in time series foundation models with our work on scientific machine learning and thoughts on science in the age of foundation models. Please find my slides here.
LLMs for Time Series Forecasting: Science in the Age of Foundation Models
Talk, School of Geography and the Environment, Oxford University, Oxford, England
I spoke at the School of Geography and the Environment at Oxford University on foundation models for time series forecasting and their potential for scientific applications related to hydro-climatic extremes. Please find my slides here.
Mitra: Tabular Foundation Model via Synthetic Pretraining
Talk, Machine Learning University (MLU), Amazon Web Services (AWS), Palo Alto, CA
I spoke at the Machine Learning University (MLU) on our custom tabular foundation model (TFM), Mitra, which is trained purely on synthetic data. We released Mitra in AutoGluon 1.4. See the tutorial on how to call Mitra with only 3 lines of code in AutoGluon 1.4. We have also released the weights on HuggingFace. Find the Mitra classifier here and regressor here.
Science in the Age of Foundation Models
Talk, Columbia Water Resource Research Seminar, New York, New York
I spoke at the Columbia Water Resource Research Seminar on recent time series foundation models including Chronos and their applications to scientific disciplines, highlighting the importance of uncertainty quantification and physical constraints. Please find my slides here.
LLMs for Time Series
Talk, Deep Learning for Science School, Lawrence Berkeley Laboratory, Berkeley, CA
I will be giving an invited talk and tutorial at the Deep Learning for Science Summer School at the Lawrence Berkeley Laboratory (LBL) on our work on foundation model for time series forecasting (Chronos) and AI4Science works. Please find the recording here.
Physics-constrained Machine Learning for PDEs
Talk, Computational Math in Industry and Beyond Seminar Series (CME 500), Stanford, CA
I will be giving an invited talk at the Computational Math in Industry and Beyond Seminar Series (CME 500) at the Institute for Computational and Mathematical Engineering (ICME) at Stanford University.
Physics-Constrained Machine Learning for Scientific Computing
Talk, Simon Fraser University (SFU) Statistics Seminar, Vancouver, Canada
I will be giving an invited talk at the Simon Fraser University (SFU) Statistics graduate seminar.
Intersection of Foundation Models and Scientific Computing
Talk, Foundation Models for Science: Progress, Opportunities, and Challenges, NeurIPS 2024 Workshop, Vancouver, Canada
I spoke at the Foundation Models for Science: Progress, Opportunities, and Challenges at NeurIPS 2024 Workshop. Please find my talk here.
ICME’s 20th Anniversary Celebration Event
Talk, Institute of Computational and Mathematical Engineering (ICME), Stanford University
In this panel, I discussed my PhD experiences at ICME and how that prepared me for a research career in industry. See the recording here and news article here.
UC Berkeley Math Career Panel
Talk, UC Berkeley Mathematics Department, University of California, Berkeley
In this panel, I discussed my journey and career in mathematics after graduating from UC Berkeley with my bachelors in Applied Mathematics and then my PhD in Computational and Mathematical Engineering from Stanford University. I advised aspiring young mathematics students on the vast career opportunities with a mathematics degree.
Physics-constrained Machine Learning for Scientific Computing
Talk, Ansys Monthly Seminar, Paris, France
I spoke at the Ansys Monthly Seminar on our physics-constrained machine learning works with applications to weather and climate forecasting and aerodyanmics.
Advances in Scientific Machine Learning (SciML) in Industry
Talk, ICERM Workshop on The Industrialization of SciML, Brown University
I gave an invited talk at the Institute for Computational and Experimental Research in Mathematics (ICERM)’s Workshop on the Industrialization of SciML at Brown University.
Physics-constrained Machine Learning for Scientific Computing
Talk, CS294 Graduate Course on Physics Inspired Deep Learning, University of California, Berkeley
I gave a guest lecture in Professor Krishnapriyan’s advanced Computer Science 294 graduate course on Physics Inspired Deep Learning at the University of California, Berkeley.
Physics-constrained Machine Learning for Earth and Sustainability Science
Talk, 2nd AI for Good Webinar Series: AI for Earth and Sustainability Science, Virtual
My talk on “Physics-constrained Machine Learning for Earth and Sustainability Science” at the 2nd AI for Good Webinar Series for AI for Earth and Sustainability Science covers our following four research works in various scientific disciplines from epidemiology to weather and climate:
- Bridging Physics-based and Data-driven modeling for Learning Dynamical Systems, L4DC, 2021.
- Learning Physical Models that Can Respect Conservation Laws, Physica D: Nonlinear Phenomena, 2024, ICML, 2023.
- Guiding Continuous Operator Learning through Physics-based boundary constraints, ICLR, 2023.
- PreDiff: Precipitation Nowcasting with Latent Diffusion Models, NeurIPS, 2023.
Women in Science at Amazon: A Conversation with our Amazonians Panelists
Talk, Amazon Science, Virtual
In this panel, we discuss challenges for women in STEM and how to persevere, internship mentoring opportunities and the future uses of machine learning in science.
Physics-constrained Machine Learning for Scientific Computing
Talk, Machine Learning and Dynamical Systems Seminar, Alan Turing Institute, London, England
My invited talk on “Physics-constrained Machine Learning for Scientific Computing” at the Machine Learning and Dynamical Systems Seminar at the Alan Turing Institute covers our following three research works:
Mathematics in Science
Talk, Daves Avenue Elementary School Science Fair, Los Gatos, CA
Please find my talk here.
The Intersection of Computational Math and Industry
Talk, Computational Math in Industry and Beyond Seminar Series (CME 500), Stanford, CA
Neural Time Series Models with GluonTS
Talk, International Conference on Machine Learning (ICML) Time Series Workshop, Long Beach, CA
ICML Time Series Workshop Talk with corresponding youtube link on GluonTS.
Temporal oscillations in the porous medium equation: why harmonic averaging itself is not to blame
Talk, SIAM Conference on Mathematical and Computational Issues in the Geosciences 2017, Erlangen, Germany
PhD Thesis Research on finite-volume averaged-based methods for nonlinear porous media flow
ICME Xpo Research Symposium
Talk, Institute for Computational and Mathematical Engineering, Stanford University, Stanford, CA
I presented posters on my research at the ICME Xpo Research Symposium from 2015-2018 on the following topics:
- Numerical Artifacts in the Generalized Porous Medium Equation and Solutions, PhD Thesis Research, 2017-2018
- Sparse Matrix Vector Multiplication Using the Merge Path, NVIDIA, 2016
- Won the ICME Xpo Best Poster Presenter Award
- Applications of the Voronoi Implicit Interface Method for Shape Optimization Problems Involving Interconnected Regions, Lawrence Berkeley National Laboratory, 2015