Approximation Algorithms for Cascading Prediction Models
Abstract
We present an approximation algorithm that takes a pool of pre-trained models as input and produces from it a cascaded model with similar accuracy but lower average-case cost. Applied to state-of-the-art ImageNet classification models, this yields up to a 2x reduction in floating point multiplications, and up to a 6x reduction in average-case memory I/O. The auto-generated cascades exhibit intuitive properties, such as using lower-resolution input for easier images and requiring higher prediction confidence when using a computationally cheaper model.
Cite
Text
Streeter. "Approximation Algorithms for Cascading Prediction Models." International Conference on Machine Learning, 2018.Markdown
[Streeter. "Approximation Algorithms for Cascading Prediction Models." International Conference on Machine Learning, 2018.](https://mlanthology.org/icml/2018/streeter2018icml-approximation/)BibTeX
@inproceedings{streeter2018icml-approximation,
title = {{Approximation Algorithms for Cascading Prediction Models}},
author = {Streeter, Matthew},
booktitle = {International Conference on Machine Learning},
year = {2018},
pages = {4752-4760},
volume = {80},
url = {https://mlanthology.org/icml/2018/streeter2018icml-approximation/}
}