Machine Learning for Probabilistic and Conformal Prediction Workshop

Event Overview:

While basic uncertainty quantification provides valuable insights, it often falls short when it comes to capturing the intricacies of real world prediction tasks such as diagnosing illnesses. Conformal prediction provides rigorous uncertainty claims based on the assumption that the data is exchangeable; it does not assume which predictive algorithm was used.

The workshop will begin with our speaker, Valeriy Manokhin, presenting a basic overview of these methods and their use-cases. Key topics include validity guarantees, calibration properties, and predictive performance evaluation within the conformal prediction framework. Participants are encouraged to ask questions throughout the presentation. After any lingering questions have been answered, Veronika Polushina will start a simple step-by-step walk through on how to leverage this tool using Pythons scikit learn library.

Links:

Slides: www.researchgate.net/publication/371566526_Machine…
GitHub code for tutorial: github.com/MLWorkshops/MAPIEwalkthrough
Google colab for tutorial: colab.research.google.com/drive/1NfnUL1bTOvZRmVYHV…
Conformal Prediction Slack Channel: join.slack.com/t/awesomeconformalpred/shared_invit…
Resources for Valeriy's upcoming book: github.com/PacktPublishing/Practical-Guide-to-Appl…

Event hosted by: www.meetup.com/Seattle-Artificial-Intelligence-Wor…