Neural Dynamic Programming for Musical Self Similarity
Abstract
We present a neural sequence model designed specifically for symbolic music. The model is based on a learned edit distance mechanism which generalises a classic recursion from computer science, leading to a neural dynamic program. Repeated motifs are detected by learning the transformations between them. We represent the arising computational dependencies using a novel data structure, the edit tree; this perspective suggests natural approximations which afford the scaling up of our otherwise cubic time algorithm. We demonstrate our model on real and synthetic data; in all cases it out-performs a strong stacked long short-term memory benchmark.
Cite
Text
Walder and Kim. "Neural Dynamic Programming for Musical Self Similarity." International Conference on Machine Learning, 2018.Markdown
[Walder and Kim. "Neural Dynamic Programming for Musical Self Similarity." International Conference on Machine Learning, 2018.](https://mlanthology.org/icml/2018/walder2018icml-neural/)BibTeX
@inproceedings{walder2018icml-neural,
title = {{Neural Dynamic Programming for Musical Self Similarity}},
author = {Walder, Christian and Kim, Dongwoo},
booktitle = {International Conference on Machine Learning},
year = {2018},
pages = {5105-5113},
volume = {80},
url = {https://mlanthology.org/icml/2018/walder2018icml-neural/}
}