Learning Stable Deep Dynamics Models

Authors: J. Zico Kolter, Gaurav Manek

NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental 4 Empirical results We illustrate our technique on several example problems, first highlighting the (inherent) stability of the method for random networks, demonstrating learning on simple n-link pendulum dynamics, and finally learning high-dimensional stable latent space dynamics for dynamic video textures via a VAE model.
Researcher Affiliation Collaboration Gaurav Manek Department of Computer Science Carnegie Mellon University gmanek@cs.cmu.edu J. Zico Kolter Department of Computer Science Carnegie Mellon University and Bosch Center for AI zkolter@cs.cmu.edu
Pseudocode No The paper contains mathematical equations and functional descriptions but no clearly labeled 'Pseudocode' or 'Algorithm' blocks.
Open Source Code No The paper does not contain any explicit statements or links indicating that the source code for the described methodology is publicly available.
Open Datasets No The paper mentions training data for n-link pendulums is 'produced by the symbolic algebra solver sympy, using simulation code adapted from [21]' and for video textures is 'a sequence of frames sampled from videos', but does not provide concrete access information (link, DOI, formal citation) for any publicly available dataset.
Dataset Splits No The paper does not provide specific dataset split information (percentages, sample counts, or citations to predefined splits) for training, validation, and testing.
Hardware Specification No The paper does not provide specific hardware details (e.g., exact GPU/CPU models, memory amounts) used for running experiments.
Software Dependencies No The paper mentions 'Py Torch' and 'sympy' but does not provide specific version numbers for these or any other software dependencies needed to replicate the experiment.
Experiment Setup Yes Specifically, we let ˆf be defined by a 2-100-100-2 fully connected network, and V be a 2-100-100-1 ICNN, with both networks initialized via the default weights of Py Torch (the Kaiming uniform initialization [8]) and with the ICNN having it s U weights further put through a softplus unit to make them positive. We train the full system to minimize minimize e,d, ˆ f,V KL(N(µt, σ2 t I N(0, I)) + Ez d(zt) yt 2 2 + d(f(zt)) yt+1 2 2 (21)