Learning One-hidden-layer Neural Networks with Landscape Design
Authors: Rong Ge, Jason D. Lee, Tengyu Ma
ICLR 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We also prove finite sample complexity results and validate the results by simulations. |
| Researcher Affiliation | Collaboration | Rong Ge Computer Science Department Duke University rongge@cs.duke.edu Jason D. Lee Data Sciences and Operations Department, University of Southern California jasonlee@marshall.usc.edu Tengyu Ma Facebook AI Research tengyuma@cs.stanford.edu |
| Pseudocode | No | The paper describes algorithms (e.g., Algorithm 1, Algorithm 2 in Section D), but these are presented as textual descriptions rather than structured pseudocode blocks. |
| Open Source Code | No | The paper does not provide an explicit statement about releasing code for the methodology or a link to a code repository. |
| Open Datasets | No | The paper mentions data being 'generated from a one-hidden-layer network' in simulations but does not refer to a publicly available dataset with concrete access information (link, DOI, formal citation). |
| Dataset Splits | No | The paper mentions 'fresh samples at each iteration' and 'test error' but does not specify any training/test/validation dataset splits (e.g., percentages or counts). |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., exact GPU/CPU models, processor types with speeds, memory amounts, or detailed computer specifications) used for running its experiments. |
| Software Dependencies | No | The paper does not provide specific ancillary software details (e.g., library or solver names with version numbers) needed to replicate the experiment. |
| Experiment Setup | Yes | We used batch size 262144 in the experiment for G( ). However, in contrast, for the ˆσ2h2 + ˆσ4h4 we used batch size 8192 and for relu we used batch size 256. We decreased step-size by a factor of 4 every 5000 number of iterations after the error plateaus at 10000 iterations. |