Input-Cell Attention Reduces Vanishing Saliency of Recurrent Neural Networks

Authors: Aya Abdelsalam Ismail, Mohamed Gunady, Luiz Pessoa, Hector Corrada Bravo, Soheil Feizi

NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Using synthetic data, we show that the saliency map produced by the input-cell attention RNN is able to faithfully detect important features regardless of their occurrence in time. We also apply the input-cell attention RNN on a neuroscience task analyzing functional Magnetic Resonance Imaging (f MRI) data for human subjects performing a variety of tasks.
Researcher Affiliation Academia 1 Department of Computer Science, University of Maryland 2 Department of Psychology, University of Maryland {asalam,mgunady}@cs.umd.edu, pessoa@umd.edu, hcorrada@umiacs.umd.edu, sfeizi@cs.umd.edu
Pseudocode No The paper does not contain structured pseudocode or algorithm blocks.
Open Source Code Yes Code available at https://github.com/ayaabdelsalam91/Input-Cell-Attention
Open Datasets Yes We apply input-cell attention to an openly available f MRI dataset of the Human Connectome Project (HCP) [26].
Dataset Splits No The paper mentions training and testing but does not provide specific details on validation splits or percentages for any dataset.
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models, memory) used for running the experiments.
Software Dependencies No The paper does not provide specific software dependencies with version numbers.
Experiment Setup No The paper describes the datasets used and the types of experiments conducted (e.g., binary classification, training to convergence) but lacks specific hyperparameter values (e.g., learning rate, batch size, number of epochs) or detailed system-level training settings.