38. Prof. Sergei Kalinin - Machine Learning for Automated Experiments

[@elizavetabelova6525]

Wonderful talk! Can’t appreciate enough how Prof. Kalinin explains complex concepts in a very simple and understandable manner.

[@ElectrochemicalColloquium]

Thank you for watching, Elizaveta!

[@dvivek07]

Fantastic lecture. Even more than the lecture I enjoyed the question answer part. Thanks for sharing!

[@ElectrochemicalColloquium]

Thanks for watching!

[@ElectrochemicalColloquium]

00:00 - Introduction
02:28 - Career path and how ML came into play
04:20 - Opportunities in Materials and ML
16:05 - Instrumentation, Cloud Computing and User Facilities
19:10 - Workflow design and implementation
29:33 - Machine Learning for physical sciences
35:24 - Machine Learning in industry
39:49 - Data analysis using Machine Learning (in microscopy)
47:20 - Automated experiments for aspects that we know in advance
55:49 - Automated experiments via Bayesian Optimization
1:07:31 - Hypothesis Active Learning
1:14:23 - Concluding remarks and the future of the field
Q&A:
1:20:33 - Q1: Advice for newcomers
1:28:40 - Q2: Commercialization of AI-controlled instruments
1:30:18 - Q3: Machine learning for exploring complex phenomena
1:33:40 - Q4: Accidental discoveries
1:39:03 - Q5: Machine learning with small datasets (e.g. at synchrotrons)
1:44:58 - Q6: AI-assisted discovery at a scale: synthesis, characterization, etc.
1:49:55 - Q7: Discoveries by learning on large-scale datasets
1:14:49 - Q8: Is there a need for Department of Machine Learning and AI?