Explore personalized machine learning systems in this 50-minute talk by Julian McAuley, PhD. Discover the principles behind recommender systems, personalized natural language processing, and computer vision applications. Learn about traditional supervised learning, complex recommender systems, personalized language generation, and visual data modeling. Examine the technical setup, model-based vs. contextual personalization, and generative text models. Investigate practical applications like assistive writing, persona-grounded dialog, and complementary item recommendations. Delve into fairness considerations and the consequences of deploying personalized predictive systems in various domains.
Overview
Syllabus
Intro
What is Personalized Machine Learning?
"Traditional" Supervised Machine Learning
Personalized Machine Learning: Recommendation
Technical setup
The Netflix prize
More Complex Recommender Systems
Model-based vs. contextual personalization
Personalized Language Generation & Explanation
Generative models of text
Generating reviews (example)
Personalized explanations (examples)
Assistive writing and expansion
Persona-grounded dialog
Personalized models of visual data
Starting point visual models for recommendation
Complementary item recommendation
Fit prediction
Modeling personalized fitness dynamics
Conversational recommendation
Personalized Design
Fairness
Taught by
Open Data Science
