Episode 4
Charles Yang: Machine Learning for Scientific Research
Charles Yang is an EECS masters student at UC Berkeley focusing on AI and dynamical systems. He writes the excellent Machine Learning For Science newsletter where he showcases a wide range of use cases for machine learning in scientific research and engineering. Learn more about Charles:
Website: https://charlesxjyang.github.io/
Google Scholar: https://scholar.google.com/citations?user=BYOREdwAAAAJ&hl=en
ML4Sci Newsletter (Highly Recommended!): https://ml4sci.substack.com/
Want to level-up your skills in machine learning and software engineering? Subscribe to our newsletter: https://mlengineered.ck.page/943aa3fd46
Take the Giving What We Can Pledge: https://www.givingwhatwecan.org/
Subscribe to ML Engineered: https://www.mlengineered.com/listen
Follow Charlie on Twitter: https://twitter.com/CharlieYouAI
Timestamps:
(02:08) Getting started in material science and machine learning
(08:58) "ImageNet moment" for ML in science
(13:20) Model explainability and transparency
(17:06) Charles' Current Research
(18:40) Embedding existing knowledge into ML models
(22:26) "Bilingual Scientists"
(24:46) Learning ML as a traditional scientist
(28:22) Private vs Public ML Research
(32:42) Rise of open-access research
(35:22) "SOTA chasing" in ML research
(38:10) Scientific ML research processes
(44:34) Applying ML knowledge to a scientific problem
(48:00) Biggest opportunities for ML in science
(51:18) Diversity in the research community
(54:24) Writing the ML4Sci newsletter
(56:20) Keeping up with new research
(01:05:30) Rapid Fire Questions
Links:
Charles' article on AI-powered Science as a Service
Charles' article on Deep Learning in Science
Charles' article on Scientific Gatekeeping
Charles' article on Open Access Research
Google Weather Forecasting paper
Google 2nd Weather Forecasting paper
DeepMind Protein Folding paper
SalesForce Protein Folding paper
ML speeding up simulations by 9+ orders of magnitude (!)
Oak Ridge AI for Science Report
Nature paper using word2vec on MatSci papers
Paper using Graph NNs to find dark matter concentrations
