Predictive Projections

نویسنده

  • Nathan Sprague
چکیده

This paper addresses the problem of learning control policies in very high dimensional state spaces. We propose a linear dimensionality reduction algorithm that discovers predictive projections: projections in which accurate predictions of future states can be made using simple nearest neighbor style learning. The goal of this work is to extend the reach of existing reinforcement learning algorithms to domains where they would otherwise be inapplicable without extensive engineering of features. The approach is demonstrated on a synthetic pendulum balancing domain, as well as on a robot domain requiring visually guided control.

برای دانلود رایگان متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Level of Grammatical Proficiency and Acquisition of Functional Projections: The case of Iranian learners of English language

Unlike Lexical Projections, Functional Projections (Extended Projections) are more of an ‘abstract’ in nature. Therefore, Functional Projections seem to be acquired later than Lexical Projections by the L2 learners. The present study investigates Iranian L2 learners’ acquisition of English Extended Projections taking into account their level of grammatical proficiency. Specifically, the aim is ...

متن کامل

m-Projections involving Minkowski inverse and range symmetric property in Minkowski space

In this paper we study the impact of Minkowski metric matrix on a projection in the Minkowski Space M along with their basic algebraic and geometric properties.The relation between the m-projections and the Minkowski inverse of a matrix A in the minkowski space M is derived. In the remaining portion commutativity of Minkowski inverse in Minkowski Space M is analyzed in terms of m-projections as...

متن کامل

Lossless image compression via predictive coding of discrete Radon projections

This paper investigates predictive coding methods to compress images represented in the Radon domain as a set of projections. Both the correlation within and between discrete Radon projections at similar angles can be exploited to achieve lossless compression. The discrete Radon projections investigated here are those used to define the Mojette Transform first presented by Guédon et al in 1995 ...

متن کامل

Letters to the Editor RE: ‘‘BAYESIAN PROJECTIONS: WHAT ARE THE EFFECTS OF EXCLUDING DATA FROM YOUNGER AGE GROUPS?’’ Baker and Bray

Baker and Bray (1) continue Bray’s thoughtful development of a particular Bayesian age-period-cohort model for cancer rate projections. This model has an elegant formulation and has brought to the forefront a Bayesian interpretation of cancer projections. However, we feel that there are more appropriate models for cancer rate projections. Why? The Bayesian age-period-cohort model suffers from v...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2009