Reinforcing Neural Network Stability with Attractor Dynamics

Authors: Hanming Deng, Yang Hua, Tao Song, Zhengui Xue, Ruhui Ma, Neil Robertson, Haibing Guan3765-3772

AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Through intensive experiments, we show that RMAN-modified attractor dynamics bring a more structured representation space to Res Net and its variants, and more importantly improve the generalization ability of Res Net-like networks in supervised tasks due to reinforced stability.
Researcher Affiliation Academia 1Shanghai Jiao Tong University, {denghanmig, songt333, zhenguixue, ruhuima, hbguan}@sjtu.edu.cn 2Queen s University Belfast, {Y.Hua, N.Robertson}@qub.ac.uk
Pseudocode No The paper includes mathematical equations (e.g., Eq. 1-6) but does not present any structured pseudocode or algorithm blocks.
Open Source Code No The paper does not contain any explicit statements or links indicating that the source code for the described methodology is publicly available.
Open Datasets Yes We conduct various experiments on the datasets for various tasks, including CIFAR10, CIFAR100 (Krizhevsky and Hinton 2009) for image classification and Google Commands (Warden 2017) for audio classification.
Dataset Splits No The paper describes data augmentation and training parameters, but does not provide explicit numerical details on dataset splits (e.g., percentages or sample counts) for training, validation, and testing.
Hardware Specification No The paper does not provide specific details about the hardware used to run the experiments (e.g., CPU, GPU models, memory specifications).
Software Dependencies No The paper mentions using SGD for training but does not provide specific software dependencies or version numbers (e.g., programming languages, libraries, frameworks).
Experiment Setup Yes We train all the networks using SGD with a batch size of 128 and momentum of 0.9. The learning rate starts with 0.1 and is devided by 10 at epoch 80/150, 120/225 and training terminates at epoch 160/300 on CIFAR10/CIFAR100, respectively. We apply a weight decay of 0.0005 for all networks.