@inproceedings{5ac2cf56b1a94ba28e866192aeaf16ed,
title = "Mutual Information Decay Curves and Hyper-parameter Grid Search Design for Recurrent Neural Architectures",
abstract = "We present an approach to design the grid searches for hyper-parameter optimization for recurrent neural architectures. The basis for this approach is the use of mutual information to analyze long distance dependencies (LDDs) within a dataset. We also report a set of experiments that demonstrate how using this approach, we obtain state-of-the-art results for DilatedRNNs across a range of benchmark datasets.",
keywords = "Hyper-parameter tuning, Long Distance Dependencies, Recurrent neural architectures, Vanishing gradients",
author = "Abhijit Mahalunkar and Kelleher, \{John D.\}",
note = "Publisher Copyright: {\textcopyright} 2020, Springer Nature Switzerland AG.; 27th International Conference on Neural Information Processing, ICONIP 2020 ; Conference date: 18-11-2020 Through 22-11-2020",
year = "2020",
doi = "10.1007/978-3-030-63823-8\_70",
language = "English",
isbn = "9783030638221",
series = "Communications in Computer and Information Science",
publisher = "Springer Science and Business Media Deutschland GmbH",
pages = "616--624",
editor = "Haiqin Yang and Kitsuchart Pasupa and Leung, \{Andrew Chi-Sing\} and Kwok, \{James T.\} and Chan, \{Jonathan H.\} and Irwin King",
booktitle = "Neural Information Processing - 27th International Conference, ICONIP 2020, Proceedings",
address = "Germany",
}