Stochastic Gradient Descent-Induced Drift of Representation in a Two-Layer Neural Network

Authors: Farhad Pashakhanloo, Alexei Koulakov

ICML 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Motivated by recent experimental findings of stimulus-dependent drift in the piriform cortex, we use theory and simulations to study this phenomenon in a two-layer linear feedforward network.
Researcher Affiliation Academia 1Cold Spring Harbor Laboratory, Cold Spring Harbor, New York, NY, USA.
Pseudocode No The paper contains mathematical derivations and theoretical models but no pseudocode or algorithm blocks.
Open Source Code No The paper does not contain any statement or link indicating the availability of open-source code for the described methodology.
Open Datasets No The paper describes generating synthetic data ('stimuli are drawn randomly and independently from a standard n-dimensional Gaussian distribution') but does not refer to a publicly available or open dataset with concrete access information.
Dataset Splits No The paper discusses numerical simulations and analytical derivations but does not specify dataset splits (e.g., training, validation, test percentages or counts) in the context of machine learning datasets.
Hardware Specification No The paper does not provide any specific details about the hardware used to run the experiments (e.g., GPU/CPU models, memory).
Software Dependencies No The paper does not specify any software dependencies with version numbers.
Experiment Setup Yes In both plots m = n = 10, p = 20, γ = 0.04, and η = 0.005. (bottom) History of representations for three trial stimuli after 2.2 105 training steps. n = p = 3, γ = 0.1, η = 0.1, and α = 0.5.