Wed 3 June 2020 at 12:15pm U.S. Mountain time:
Statistical Workflow and the Fractal Nature of Scientific Revolutions
How would an A.I. do statistics? Fitting a model is the easy part. The other steps of workflow—model building, checking, and revision—are not so clearly algorithmic. It could be fruitful to simultaneously think about automated inference and best practices in statistical workflow, moving beyond inference for a single model toward statistical problem solving.
These same issues arise when a human research team designs and analyzes quantitative data. We discuss in the context of examples from many different application areas.
This article from PLoS One by Cawi, Nehorai and La Rosa is a TDA approach to workflows.
Designing machine learning workflows with an application to topological data analysis
https://www.researchgate.net/publication/337692031_Designing_machine_learning_workflows_with_an_application_to_topological_data_analysis
That’s about an hour from me. Looks like it’s virtual. Very cool.
Nice topic!
Looks like it’s a private event, is it possible to register and/or have the recording afterwards?
I agree with Federico, will this be available electronically somehow?
It’ll be streamed live on Facebook!
https://www.facebook.com/events/259115335298218
The title is reminiscent of Andrew Abbott’s book about the Chaos of Disciplines. Is there a connection?
It’s a room full of Andrews here . . . I have not actually read Abbott’s book! Maybe I should.
Is model building usually thought of as statistics? It sounds more to me that they describe an AI for conducting science. I would never expect any statistical methodology to invent new hypotheses for me. That’s an altogether separate concern.
The general idea brings to mind Jaynes’ robot in the beginning chapters of LoS.
Of course it’s unclear to you how you could design an algorithm for workflow–that’s what a workflow-algorithm generating algorithm is for! The code for which will be written by a workflow-algorithm-generator generator algorithm…