A Physics-Based Model Prior for Object-Oriented MDPs

Authors: Jonathan Scholz, Martin Levihn, Charles Isbell, David Wingate

ICML 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our results show that this representation can result in much faster learning, by virtue of its strong but appropriate inductive bias in physical environments. We evaluate OO-LWR and PBRL w.r.t noise sensitivity, scalability, and model mis-specification (PBRL only).
Researcher Affiliation Academia Jonathan Scholz JKSCHOLZ@GATECH.EDU Martin Levihn LEVIHN@GATECH.EDU Charles L. Isbell ISBELL@GATECH.EDU Georgia Institute of Technology, 801 Atlantic Dr. Atlanta, GA 30332 USA David Wingate WINGATED@MIT.EDU Massachusetts Institute of Technology, 77 Massachusetts Ave., Cambridge, MA 02139 USA
Pseudocode No The paper does not contain structured pseudocode or algorithm blocks.
Open Source Code No The paper does not provide concrete access to source code for the methodology described in this paper. It mentions using “Box2D engine (Catto, 2013). URL http://www.box2d.org” which is a third-party tool.
Open Datasets No The paper describes custom simulated environments (“Shopping Cart Task” and “Apartment Rearrangement Task”) but does not provide concrete access information (link, DOI, repository, or formal citation with authors/year) for a publicly available or open dataset.
Dataset Splits No The paper discusses “training observations” but does not provide specific dataset split information (exact percentages, sample counts, citations to predefined splits, or detailed splitting methodology) needed to reproduce the data partitioning.
Hardware Specification No The paper does not provide specific hardware details (exact GPU/CPU models, processor types with speeds, memory amounts, or detailed computer specifications) used for running its experiments.
Software Dependencies No The paper mentions “Box2D engine (Catto, 2013)” but does not provide a specific version number for this or any other software dependency.
Experiment Setup Yes Table 2. Table of the relevant algorithm parameters for each experiment. k: number of nearest neighbors (LWR,OO-LWR), λ: bandwidth (LWR,OO-LWR), ns: number of sectors (OO-LWR), ntps: number of raycast collision tests per sector (OO-LWR), ϵc: collision radius (OO-LWR), prior: type of prior (PBRL), MCMC: sampler parameters (iterations, burn-in, thin, number of chains) (PBRL). (followed by specific values for each task)