Online Prediction in Sub-linear Space
From Google TechTalks · 1:06:16
Show by Google TechTalks
A Google TechTalk, presented by Fred Zhang, 2023/06/06Google Algorithms Seminar. ABSTRACT: We design the first sub-linear memory algorithm for online learning with expert advice, arguably the most basic question in online and sequential decision making. This problem is solved classically by the well-known multiplicative weights update method, which achieves optimal regret but suffers a linear space complexity. We show how to bypass this barrier. In this talk, I will discuss the main techniques, recent followup works, and many open directions. Joint work with Binghui Peng (https://arxiv.org/abs/2207.07974, SODA 23).Uploaded 11-06-23About the speaker: Fred Zhang is a fifth-year PhD student in the theory group at Berkeley, advised by Jelani Nelson. He is broadly interested in algorithmic questions in learning and statistics.
More episodes of Google TechTalks
More episodes
- Threat Models for Memorization: Privacy, Copyright, and Everything In-Between
- Technology Assisted Reading Acquisition: Children Acquiring Literacy Naturally
- sex on the internet, the realities of porn, sexual privacy,
- Privacy Auditing of Large Language Models
- The Limits and Possibilities of One Run Auditing
- PhotoTechEDU Day 29: Photographing VR Panoramas
- Privacy Ripple Effects from Adding or Removing Personal Information in Language Model Training
- Inside VMware Fusion
- Java Posse Episode 100 (Live Special)
- Cascading Adversarial Bias from Injection to Distillation in Language Models
PodSearch is a free, independent podcast search engine. Read about PodSearch, the privacy policy or contact us.