Neural Functional Programming
Abstract
We discuss a range of modeling choices that arise when constructing an end-to-end differentiable programming language suitable for learning programs from input-output examples. Taking cues from programming languages research, we study the effect of memory allocation schemes, immutable data, type systems, and built-in control-flow structures on the success rate of learning algorithms. We build a range of models leading up to a simple differentiable functional programming language. Our empirical evaluation shows that this language allows to learn far more programs than existing baselines.
Cite
Text
Feser et al. "Neural Functional Programming." International Conference on Learning Representations, 2017.Markdown
[Feser et al. "Neural Functional Programming." International Conference on Learning Representations, 2017.](https://mlanthology.org/iclr/2017/feser2017iclr-neural/)BibTeX
@inproceedings{feser2017iclr-neural,
title = {{Neural Functional Programming}},
author = {Feser, John K. and Brockschmidt, Marc and Gaunt, Alexander L. and Tarlow, Daniel},
booktitle = {International Conference on Learning Representations},
year = {2017},
url = {https://mlanthology.org/iclr/2017/feser2017iclr-neural/}
}