README.md

Sequence Autoencoder with UJI Pen Characters Dataset

Unofficial TensorFlow implementation of Sequence Autoencoder with UJI Pen Characters Dataset.
The Keras implementation is provided as the following link.
https://github.com/kjm1559/lstm_autoencoder

Usage

$ unzip dataset_npz.zip
$ python run.py

Dataset

Sample view of UJI Pen Characters V1 [1]. The left figure uses only one pen stroke and the right one uses two pen strokes.

Result

Neural network architecture referenced from DeVries et al. [2].

Result

Loss graph during the training procedure.

Generated Pen Characters from test data.

Reference

[1] Dua, D. and Graff, C. (2019). UCI Machine Learning Repository [http://archive.ics.uci.edu/ml]. Irvine, CA: University of California, School of Information and Computer Science.
[2] DeVries, Terrance, and Graham W. Taylor. "Dataset augmentation in feature space." arXiv preprint arXiv:1702.05538 (2017).