How regularization affects the critical points in linear networks
Authors: Amirhossein Taghvaei, Jin W. Kim, Prashant Mehta
NeurIPS 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Analytical and numerical tools from bifurcation theory are used to compute the critical points via the solutions of the characteristic equation. Figure 1: (a) Critical points in Example 1 (the (2, 1) entry of the solution matrix C(λ; n) is depicted for n = 0, 1, 2); (b) The cost J[A] for these solutions. Figure 2: (a) Numerical continuation of the solution in Example 2; (b) The cost J[A] for the critical point (minimum) and the constant 1/T log(R) solution. |
| Researcher Affiliation | Academia | Amirhossein Taghvaei Coordinated Science Laboratory University of Illinois at Urbana-Champaign Urbana, IL, 61801 taghvae2@illinois.edu; Jin W. Kim Coordinated Science Laboratory University of Illinois at Urbana-Champaign Urbana, IL, 61801 jkim684@illinois.edu; Prashant G. Mehta Coordinated Science Laboratory University of Illinois at Urbana-Champaign Urbana, IL, 61801 mehtapg@illinois.edu |
| 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 'Py DSTool Clewley et al. [2007]' but does not state that its own code will be released. |
| Open Datasets | No | The paper describes a theoretical model and uses 'i.i.d. input-output samples' for learning, but it does not specify or provide access information for a publicly available dataset. |
| Dataset Splits | No | The paper does not specify dataset splits for training, validation, or testing, as it focuses on theoretical analysis and numerical examples rather than empirical evaluation on a specific dataset. |
| Hardware Specification | No | The paper mentions numerical computations but does not provide any specific hardware details such as GPU/CPU models, memory, or cloud resources used for running experiments. |
| Software Dependencies | Yes | The software package Py DSTool Clewley et al. [2007] is used to numerically continue the solution C(λ; n) as a function of the parameter λ. |
| Experiment Setup | No | The paper focuses on theoretical derivations and numerical examples with specific matrix parameters, but it does not provide details on experimental setup parameters such as learning rates, batch sizes, epochs, or model initialization for a learning algorithm. |