The Dynamics of Learning: A Random Matrix Approach

Authors: Zhenyu Liao, Romain Couillet

ICML 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our results provide rich insights into common questions in neural nets, such as overfitting, early stopping and the initialization of training, thereby opening the door for future studies of more elaborate structures and models appearing in today s neural networks. ... We close this article with experiments on the popular MNIST dataset (Le Cun et al., 1998) (number 1 and 7). ... We observe an extremely close fit between our results and the empirical simulations, as shown in Figure 6.
Researcher Affiliation Academia 1Laboratoire des Signaux et Syst emes (L2S), Centrale Sup elec, Universit e Paris-Saclay, France; 2G-STATS Data Science Chair, GIPSA-lab, University Grenobles-Alpes, France.
Pseudocode No The paper does not contain any structured pseudocode or algorithm blocks.
Open Source Code No The paper does not provide concrete access to source code for the methodology described.
Open Datasets Yes We close this article with experiments on the popular MNIST dataset (Le Cun et al., 1998) (number 1 and 7).
Dataset Splits No The paper mentions selecting 'training sets' and evaluating 'generalization performance' on 'unseen data', but does not explicitly describe a validation set or specific percentages/counts for train/validation/test splits.
Hardware Specification No The paper does not provide specific hardware details (e.g., exact GPU/CPU models, 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 Figure 1. Training and generalization performance for µ = [2; 0p 1], p = 256, n = 512, σ2 = 0.1, = 0.01 and c1 = c2 = 1/2. ... Figure 6. Training and generalization performance for MNIST data (number 1 and 7) with n = p = 784, c1 = c2 = 1/2, = 0.01 and σ2 = 0.1.