Benefits of Overparameterized Convolutional Residual Networks: Function Approximation under Smoothness Constraint

Authors: Hao Liu, Minshuo Chen, Siawpeng Er, Wenjing Liao, Tong Zhang, Tuo Zhao

ICML 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Numerical experiments on adversarial robust image classification are provided to support our theory. We verify our theory by numerical experiments.
Researcher Affiliation Collaboration 1Department of Mathematics, Hong Kong Baptist University, Kowloon Tong, Kowloon, Hong Kong. 2School of Industrial and Systems Engineering, Georgia Institute of Technology, Atlanta, GA 30332 USA. 3School of Mathematics, Georgia Institute of Technology, Atlanta, GA 30332 USA. 4Department of Mathematics and Department of Computer Science and Engineering, The Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong. 5Google Research.
Pseudocode No The paper describes methods mathematically and through diagrams, but does not contain any structured pseudocode or algorithm blocks.
Open Source Code No The paper does not contain any explicit statement about releasing source code or provide links to a code repository.
Open Datasets Yes We use the CIFAR-10 data set.
Dataset Splits No The paper mentions training and test datasets but does not explicitly specify train/validation/test splits or their proportions.
Hardware Specification No The paper describes the training process and hyperparameters, but does not specify any particular hardware (e.g., GPU model, CPU type, memory) used for running the experiments.
Software Dependencies No The paper discusses various models and attack methods, but does not list specific software libraries or their version numbers used in the experiments.
Experiment Setup Yes Hyperparameters in training are set as follows: perturbation diameter ϵ = 0.031 under the ℓ∞ norm, step size for generating perturbation 0.007, number of iterations 10, learning rate 0.1, batch size b = 128 and run 76 epochs on the training dataset.