|
View: |
Part 1: Document Description
|
|
Citation |
|
|---|---|
|
Title: |
Visual Analytics System for Hidden States in Recurrent Neural Networks |
|
Identification Number: |
doi:10.18419/darus-2052 |
|
Distributor: |
DaRUS |
|
Date of Distribution: |
2021-09-10 |
|
Version: |
1 |
|
Bibliographic Citation: |
Munz, Tanja; Garcia, Rafael; Weiskopf, Daniel, 2021, "Visual Analytics System for Hidden States in Recurrent Neural Networks", https://doi.org/10.18419/darus-2052, DaRUS, V1 |
|
Citation |
|
|
Title: |
Visual Analytics System for Hidden States in Recurrent Neural Networks |
|
Identification Number: |
doi:10.18419/darus-2052 |
|
Authoring Entity: |
Munz, Tanja (University of Stuttgart) |
|
Garcia, Rafael (University of Stuttgart) |
|
|
Weiskopf, Daniel (University of Stuttgart) |
|
|
Grant Number: |
EXC 2075 - 390740016 |
|
Grant Number: |
TRR 161 - 251654672 |
|
Distributor: |
DaRUS |
|
Access Authority: |
Munz, Tanja |
|
Depositor: |
Munz, Tanja |
|
Date of Deposit: |
2021-06-27 |
|
Holdings Information: |
https://doi.org/10.18419/darus-2052 |
|
Study Scope |
|
|
Keywords: |
Computer and Information Science, Visual Analytics, Visualization, Machine Learning, Classification, Recurrent Neural Network, Long Short-Term Memory, Hidden States, Natural Language Processing, Nonlinear Projection |
|
Abstract: |
Source code of our visual analytics system for the interpretation of hidden states in recurrent neural networks.<br> This project contains source code for preprocessing data and the visual analytics system. Additionally, we added precomputed data for immediate use in the visual analysis system.<br> <br> The sub directories contain the following: <ul> <li><i>dataPreparation:</i> Python scripts to prepare data for analysis. In these scripts, Long Short-Term Memory (LSTM) models are trained and data for our visual analytics system is exported.</li> <li><i>visualAnalytics:</i> The source code of our visual analytics system to explore hidden states.</li> <li><i>demonstrationData:</i> Data files for the use with our visual analytics system. The same data can also be generated with the data preparation scripts.</li> </ul> <br> We provide two scripts to generate data for analysis in our visual analytics system: for the IMDB and Reuters dataset as available in Keras. The output files can then be loaded into our visual analytics system; their locations have to be specified in <i>userData.toml</i> of the visual analytics system. <br> <br> The output file of our data preparation scripts or the ones provided for demonstration can be loaded in our visual analytics system for visualization and analysis. Since we provide input files, you do not have to run the preprocessing steps and can use our visual analytics system immediately. <br> <br> Please have a look at the respective README-files for more details. |
|
Notes: |
You may find the most recent version of the source code on GitHub: <a href="https://github.com/MunzT/hiddenStatesVis">https://github.com/MunzT/hiddenStatesVis</a> <br> |
|
Methodology and Processing |
|
|
Sources Statement |
|
|
Data Sources: |
We generated test data for our visual analytics system from the IMDB [1] and Reuters [2] dataset available in Keras (<a href="https://github.com/fchollet/keras">https://github.com/fchollet/keras</a>).<br> The original data set is licensed undere Apache V2.0 (<a href="https://github.com/keras-team/keras/blob/master/LICENSE">https://github.com/keras-team/keras/blob/master/LICENSE</a>).<br> We processed tha data with the scripts available in the directory dataPreparation. The data can also be reproduced with the provided source code. <br> <br> [1] Maas, A.L., Daly, R.E., Pham, P.T., Huang, D., Ng, A.Y., Potts, C.: Learning word vectors for sentiment analysis. In: Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies - Volume 1. HLT’11, pp. 142-150. Association for Computational Linguistics, USA. <a href="https://dl.acm.org/doi/10.5555/2002472.2002491">https://dl.acm.org/doi/10.5555/2002472.2002491</a> (2011) <br> [2] Reuters Ltd.: Reuters-21578 Dataset. <a href="http://kdd.ics.uci.edu/databases/reuters21578/reuters21578.html">http://kdd.ics.uci.edu/databases/reuters21578/reuters21578.html</a> (1996) |
|
Origins of Sources: |
IMDB and Reuters datasets:<br> F. Chollet. Keras. GitHub. <a href="https://github.com/fchollet/keras">https://github.com/fchollet/keras</a>. 2015. |
|
Data Access |
|
|
Other Study Description Materials |
|
|
Related Publications |
|
|
Citation |
|
|
Title: |
R. Garcia, T. Munz, and D. Weiskopf. "Visual Analytics Tool for the Interpretation of Hidden States in Recurrent Neural Networks". Visual Computing for Industry, Biomedicine, and Art (VCIBA). 2021. |
|
Identification Number: |
10.1186/s42492-021-00090-0 |
|
Bibliographic Citation: |
R. Garcia, T. Munz, and D. Weiskopf. "Visual Analytics Tool for the Interpretation of Hidden States in Recurrent Neural Networks". Visual Computing for Industry, Biomedicine, and Art (VCIBA). 2021. |
|
Label: |
hiddenStatesVis.zip |
|
Text: |
Source code of our approach. |
|
Notes: |
application/zip |