Online Convex Optimization in Presence of Concept Drift

Please enable javascript in your browser.

Rasouli, Sina | 2018

1301 Viewed

Type of Document: M.Sc. Thesis
Language: Farsi
Document No: 52242 (02)
University: Sharif University of Technology
Department: Mathematical Sciences
Advisor(s): Razvan, Mohammad Reza; Alishahi, Kasra
Abstract:
The problem of learning using high volume of data as stream, has attracted much attention recently. In this thesis, the problem is modeled and analized using Online Convex Optimization tools [1], [2]. General performance bounds are stated and clarified in this framework [8]. Using the practical experience in Online Decision Making (e.g., predicting price in Stock Market), the need for a more flexible model, which adapts to changes in problem, is presented. In this thesis, after reviewing the literature and online convex optimization framework, we will define ”Concept Drift”, which describes changes in the dynamics of the problem and the statistical tools to detect it [13], [5]. And finally, we will compare and analyze the performance of online convex optimization methods, in presence of concept drift.All analysis are carried out both theoretically and using simulations
Keywords:
Online Optimization ; Convex Optimization ; Online Learning ; Concept Drift