Visual Ensemble Analysis to Study the Influence of Hyper-parameters on Training Deep Neural Networks
| dc.contributor.author | Hamid, Sagad | en_US |
| dc.contributor.author | Derstroff, Adrian | en_US |
| dc.contributor.author | Klemm, Sören | en_US |
| dc.contributor.author | Ngo, Quynh Quang | en_US |
| dc.contributor.author | Jiang, Xiaoyi | en_US |
| dc.contributor.author | Linsen, Lars | en_US |
| dc.contributor.editor | Archambault, Daniel and Nabney, Ian and Peltonen, Jaakko | en_US |
| dc.date.accessioned | 2019-06-02T18:23:44Z | |
| dc.date.available | 2019-06-02T18:23:44Z | |
| dc.date.issued | 2019 | |
| dc.description.abstract | A good deep neural network design allows for efficient training and high accuracy. The training step requires a suitable choice of several hyper-parameters. Limited knowledge exists on how the hyper-parameters impact the training process, what is the interplay of multiple hyper-parameters, and what is the interrelation of hyper-parameters and network topology. In this paper, we present a structured analysis towards these goals by investigating an ensemble of training runs.We propose a visual ensemble analysis based on hyper-parameter space visualizations, performance visualizations, and visualizing correlations of topological structures. As a proof of concept, we apply our approach to deep convolutional neural networks. | en_US |
| dc.description.sectionheaders | Papers | |
| dc.description.seriesinformation | Machine Learning Methods in Visualisation for Big Data | |
| dc.identifier.doi | 10.2312/mlvis.20191160 | |
| dc.identifier.isbn | 978-3-03868-089-5 | |
| dc.identifier.pages | 19-23 | |
| dc.identifier.uri | https://doi.org/10.2312/mlvis.20191160 | |
| dc.identifier.uri | https://diglib.eg.org:443/handle/10.2312/mlvis20191160 | |
| dc.publisher | The Eurographics Association | en_US |
| dc.title | Visual Ensemble Analysis to Study the Influence of Hyper-parameters on Training Deep Neural Networks | en_US |