EE Student Information

The Department of Electrical Engineering supports Black Lives Matter. Read more.

• • • • •

EE Student Information, Spring Quarter through Academic Year 2020-2021: FAQs and Updated EE Course List.

Updates will be posted on this page, as well as emailed to the EE student mail list.

Please see Stanford University Health Alerts for course and travel updates.

As always, use your best judgement and consider your own and others' well-being at all times.

Statistics Seminar presents "Some recent insights on transfer-learning"

Topic: 
Some recent insights on transfer-learning
Tuesday, September 22, 2020 - 4:30pm
Venue: 
Zoom ID 973 5368 4241 (+password)
Speaker: 
Samory Kpotufe (Columbia University)
Abstract / Description: 

A common situation in Machine Learning is one where training data is not fully representative of a target population due to bias in the sampling mechanism or due to prohibitive target sampling costs. In such situations, we aim to "transfer" relevant information from the training data (a.k.a. source data) to the target application. How much information is in the source data about the target application? Would some amount of target data improve transfer? These are all practical questions that depend crucially on "how far" the source domain is from the target. However, how to properly measure "distance" between source and target domains remains largely unclear.

In this talk we will argue that much of the traditional notions of distance (e.g., KL-divergence, extensions of TV such as D_A discrepancy, density-ratios, Wasserstein distance) can yield an over-pessimistic picture of transferability. Instead, we show that some new notions of "relative dimension" between source and target (which we simply term transfer-exponents) capture a continuum from easy to hard transfer. Transfer-exponents uncover a rich set of situations where transfer is possible even at fast rates; they encode relative benefits of source and target samples, and have interesting implications for related problems such as multi-task or multi-source learning.

In particular, in the case of transfer from multiple sources, we will discuss (if time permits) a strong dichotomy between minimax and adaptive rates: no adaptive procedure exists that can achieve the same rates as minimax (oracle) procedures.

The talk is based on earlier work with Guillaume Martinet, and ongoing work with Steve Hanneke.