Gradient Descent Provably Optimizes Over-parameterized Neural Networks

Authors: Simon S. Du, Xiyu Zhai, Barnabas Poczos, Aarti Singh

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

Reproducibility Variable Result LLM Response
Research Type Experimental 5 EXPERIMENTS In this section, we use synthetic data to corroborate our theoretical findings.
Researcher Affiliation Academia Simon S. Du Machine Learning Department Carnegie Mellon University ssdu@cs.cmu.edu Xiyu Zhai Department of EECS Massachusetts Institute of Technology xiyuzhai@mit.edu Barnab as Pocz os Machine Learning Department Carnegie Mellon University bapozos@cs.cmu.edu Aarti Singh Machine Learning Department Carnegie Mellon University aartisingh@cmu.edu
Pseudocode No The paper does not contain any sections or figures explicitly labeled as 'Pseudocode' or 'Algorithm', nor are there structured steps formatted like code.
Open Source Code No The paper does not contain any statement about releasing source code for the described methodology, nor does it provide any links to a code repository.
Open Datasets No We use synthetic data to corroborate our theoretical findings. We uniformly generate n = 1000 data points from a d = 1000 dimensional unit sphere and generate labels from a one-dimensional standard Gaussian distribution.
Dataset Splits No The paper states 'We use synthetic data to corroborate our theoretical findings.' but does not specify any dataset splits (training, validation, test) or cross-validation setup.
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models, memory, or cloud instances) used for running its experiments.
Software Dependencies No The paper does not provide specific software dependencies with version numbers.
Experiment Setup No For all experiments, we run 100 epochs of gradient descent and use a fixed step size.