Home › Categories › Technology› Http://www.cs.nyu.ed U/~mohri/mls/lecture _8.pdf

Http://www.cs.nyu.ed U/~mohri/mls/lecture _8.pdf

lecture_8.pdf

🔗 Source: cs.nyu.edu

📊 Size: 4.26 MB

📄 Pages: 48 pages

⬇️ Downloads: 80

Summary

On-line learning has no distributional assumption, uses worst-case analysis, and mixes training and test data, with performance measured by mistake model or regret. It is compared to PAC learning, which assumes a fixed distribution over time and IID. The general on-line setting involves receiving an instance, predicting, receiving a label, and incurring loss, with the objective of minimizing total loss. A specific case is prediction with expert advice, where the goal is to minimize regret, or the difference between the total loss incurred and that of the best expert.

Download File

Description

On-line learning has no distributional assumption, uses worst-case analysis, and mixes training and test data, with performance measured by mistake model or...

Technical Information

File Format: PDF
File Size: 4.26 MB
Pages: 48
Language: EN
Total Downloads: 80
Last Updated: 1 month ago

Document Overview

This PDF document about http://www.cs.nyu.ed u/~mohri/mls/lecture _8.pdf provides comprehensive information and guidance. Whether you're a beginner or advanced user, this resource offers valuable insights into http://www.cs.nyu.ed u/~mohri/mls/lecture _8.pdf.