Unsupervised Feature Learning from Temporal Data
Abstract
Current state-of-the-art classification and detection algorithms rely on supervised training. In this work we study unsupervised feature learning in the context of temporally coherent video data. We focus on feature learning from unlabeled video data, using the assumption that adjacent video frames contain semantically similar information. This assumption is exploited to train a convolutional pooling auto-encoder regularized by slowness and sparsity. We establish a connection between slow feature learning to metric learning and show that the trained encoder can be used to define a more temporally and semantically coherent metric.
Cite
Text
Goroshin et al. "Unsupervised Feature Learning from Temporal Data." International Conference on Learning Representations, 2015.Markdown
[Goroshin et al. "Unsupervised Feature Learning from Temporal Data." International Conference on Learning Representations, 2015.](https://mlanthology.org/iclr/2015/goroshin2015iclr-unsupervised/)BibTeX
@inproceedings{goroshin2015iclr-unsupervised,
title = {{Unsupervised Feature Learning from Temporal Data}},
author = {Goroshin, Ross and Bruna, Joan and Tompson, Jonathan and Eigen, David and LeCun, Yann},
booktitle = {International Conference on Learning Representations},
year = {2015},
url = {https://mlanthology.org/iclr/2015/goroshin2015iclr-unsupervised/}
}