Learning VIsual Predictive Models of Physics for Playing Billiards

Authors: Katerina Fragkiadaki, Pulkit Agrawal, Sergey Levine, Jitendra Malik

ICLR 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We demonstrate that our agent can accurately plan actions for playing a simulated billiards game, which requires pushing a ball into a target position or into collision with another ball.
Researcher Affiliation Academia Katerina Fragkiadaki Pulkit Agrawal Sergey Levine Jitendra Malik Electrical Engineering and Computer Science University of California, Berkeley (katef,pulkitag,svlevine,malik)@berkeley.edu
Pseudocode No The paper describes processes in narrative text and uses diagrams to illustrate concepts (e.g., Figure 5 for generating visual imaginations), but it does not include any formal pseudocode or algorithm blocks.
Open Source Code No The paper states 'We use the publicly available Caffe package for training our model.' (Section 3.1), referring to a third-party tool, but it does not provide an explicit statement or link for the open-sourcing of the authors' own implementation or experimental code.
Open Datasets No For model learning, we generate sequences of ball motions in a randomly sampled world configuration. ... For training, we pre-generated 10K such sequences. (The paper describes creating its own dataset but provides no information about its public availability or how to access it.)
Dataset Splits No The paper mentions training and testing on 'random worlds sampled from the same distribution as the training data' and 'worlds sampled from a different distribution', but it does not explicitly specify distinct training, validation, and test dataset splits with percentages or sample counts.
Hardware Specification No The paper does not provide any specific details about the hardware (e.g., GPU/CPU models, processor types, memory amounts, or cloud resources) used for running the experiments.
Software Dependencies No The paper states 'We use the publicly available Caffe package for training our model.' (Section 3.1). While it names a software package, it does not specify its version number, nor does it list any other software dependencies with version details.
Experiment Setup Yes The model is trained by minimizing the Euclidean loss between ground-truth and predicted object velocities for h time steps in the future. ... We weigh the loss in a manner that errors in predictions at a shorter time horizon are penalized more than predictions at a longer time horizon. This weighting is achieved using penalty weights wk = exp( k 1 4 ). ... The length of each sequence was sampled from the range [20, 200]. The length of the walls was sampled from a range of [300 pixels, 550 pixels]. Balls were of radius 25 pixels and uniform density. Force direction was uniformly sampled and the force magnitude was sampled from the range [30K Newtons, 80K Newtons]. Forces were only applied on the first frame. The size of visual glimpses is 600x600 pixels. ... For training, we pre-generated 10K such sequences. We constructed minibatches by choosing 50 random subsets of 20 consequent frames from this pre-generated dataset.