ResearchTrend.AI
  • Papers
  • Communities
  • Events
  • Blog
  • Pricing
Papers
Communities
Social Events
Terms and Conditions
Pricing
Parameter LabParameter LabTwitterGitHubLinkedInBlueskyYoutube

© 2025 ResearchTrend.AI, All rights reserved.

  1. Home
  2. Papers
  3. 1304.4806
33
2
v1v2v3v4 (latest)

Unsupervised model-free representation learning

17 April 2013
D. Ryabko
    CMLAI4TS
ArXiv (abs)PDFHTML
Abstract

Numerous control and learning problems face the situation where sequences of high-dimensional highly dependent data are available but no or little feedback is provided to the learner, which makes any inference rather challenging. To address this challenge, we formulate the following problem. Given a series of observations X0,…,XnX_0,\dots,X_nX0​,…,Xn​ coming from a large (high-dimensional) space X\mathcal XX, find a representation function fff mapping X\mathcal XX to a finite space Y\mathcal YY such that the series f(X0),…,f(Xn)f(X_0),\dots,f(X_n)f(X0​),…,f(Xn​) preserves as much information as possible about the original time-series dependence in X0,…,XnX_0,\dots,X_nX0​,…,Xn​. We show that, for stationary time series, the function fff can be selected as the one maximizing a certain information criterion that we call time-series information. Some properties of this functions are investigated, including its uniqueness and consistency of its empirical estimates. Implications for the problem of optimal control are presented.

View on arXiv
Comments on this paper