otsubo/chainervr

Chainer implementation of Networks for Learning Video Representations

Chainer Video Representation

Chainer implementation of Networks for Learning Video Representations

Contents

Unsupervised Learning of Video Representations using LSTMs

Located at models/unsupervised_videos.

Srivastava, Nitish, Elman Mansimov, and Ruslan Salakhudinov.
Unsupervised learning of video representations using lstms."
International conference on machine learning. 2015.

See https://github.com/emansim/unsupervised-videos for the original implementation.

Convolutional LSTM Network: A Machine Learning Approach for Precipitation Nowcasting

Located at models/conv_lstm.

Xingjian, S. H. I., et al.
"Convolutional LSTM network: A machine learning approach for precipitation nowcasting."
Advances in neural information processing systems. 2015.

Deep Episodic Memory: Encoding, Recalling, and Predicting Episodic Experiences for Robot Action Execution

Located at models/deep_episodic_memory.

Rothfuss, Jonas, et al.
"Deep Episodic Memory: Encoding, Recalling, and Predicting Episodic Experiences for Robot Action Execution."
arXiv preprint arXiv:1801.04134 (2018).

Install

1. Clone this repository
2. Install this package using pip

cd chainervr
pip install .

3. (Optional) If you plan to use with GPU, please install appropriate cupy package.

pip install cupy-cuda91  # for CUDA 9.1
# or
pip install cupy-cuda92  # for CUDA 9.2
# and so on.

Examples

• Reconstruction and Prediction of Moving Mnist Dataset

Awesome References

• Lecture CS231n at Stanford Univ.
• Padding of ConvNet in Tensorflow

Author

Yuki Furuta <furushchev@jsk.imi.i.u-tokyo.ac.jp>