The Benefits of Mixup for Feature Learning

Authors: Difan Zou, Yuan Cao, Yuanzhi Li, Quanquan Gu

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

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results verify our theoretical findings and demonstrate the effectiveness of the early-stopped Mixup training.
Researcher Affiliation Academia 1Department of Computer Science & Institute of Data Science, The University of Hong Kong 2Department of Statistics and Actuarial Science & Department of Mathematics, The University of Hong Kong 3Machine Learning Department, Carnegie Mellon University 4Department of Computer Science, University of California, Los Angeles.
Pseudocode No The paper describes training algorithms using mathematical equations and textual descriptions, but it does not include any clearly labeled pseudocode or algorithm blocks.
Open Source Code No The paper does not provide any explicit statement about releasing source code for the described methodology, nor does it include any links to a code repository.
Open Datasets Yes We conduct a proof-of-concept experiment on CIFAR-10 dataset.
Dataset Splits No The paper mentions using CIFAR-10 and synthetic data but does not specify any training/validation/test dataset splits (e.g., percentages, sample counts, or references to predefined splits).
Hardware Specification No The paper does not provide any specific details about the hardware (e.g., CPU, GPU models, or cloud resources) used to run the experiments.
Software Dependencies No The paper mentions using 'SGD with momentum 0.9' and 'Res Net18 model', but it does not specify version numbers for any software components, libraries, or frameworks used in the experiments.
Experiment Setup Yes For the two-layer CNN model and the training algorithm, we set network width m = 10, and conduct full-batch gradient descent with learning rate η = 0.05 and total iteration number T = 20000. ... For Res Net18 and Res Net34, we set the learning rate as 0.1; for Le Net and VGG16, we set the learning rate as 0.02 and 0.1 respectively.