Categorical Feature Compression via Submodular Optimization

Mohammadhossein Bateni, Lin Chen, Hossein Esfandiari, Thomas Fu, Vahab Mirrokni, Afshin Rostamizadeh
Proceedings of the 36th International Conference on Machine Learning, PMLR 97:515-523, 2019.

Abstract

In the era of big data, learning from categorical features with very large vocabularies (e.g., 28 million for the Criteo click prediction dataset) has become a practical challenge for machine learning researchers and practitioners. We design a highly-scalable vocabulary compression algorithm that seeks to maximize the mutual information between the compressed categorical feature and the target binary labels and we furthermore show that its solution is guaranteed to be within a $1-1/e \approx 63%$ factor of the global optimal solution. Although in some settings, entropy-based set functions are known to be submodular, this is not the case for the mutual information objective we consider (mutual information with respect to the target labels). To address this, we introduce a novel re-parametrization of the mutual information objective, which we prove is submodular, and also design a data structure to query the submodular function in amortized $O(\log n )$ time (where $n$ is the input vocabulary size). Our complete algorithm is shown to operate in $O(n \log n )$ time. Additionally, we design a distributed implementation in which the query data structure is decomposed across $O(k)$ machines such that each machine only requires $O(\frac n k)$ space, while still preserving the approximation guarantee and using only logarithmic rounds of computation. We also provide analysis of simple alternative heuristic compression methods to demonstrate they cannot achieve any approximation guarantee. Using the large-scale Criteo learning task, we demonstrate better performance in retaining mutual information and also verify competitive learning performance compared to other baseline methods.

Cite this Paper

BibTeX
@InProceedings{pmlr-v97-bateni19a, title = {Categorical Feature Compression via Submodular Optimization}, author = {Bateni, Mohammadhossein and Chen, Lin and Esfandiari, Hossein and Fu, Thomas and Mirrokni, Vahab and Rostamizadeh, Afshin}, booktitle = {Proceedings of the 36th International Conference on Machine Learning}, pages = {515--523}, year = {2019}, editor = {Chaudhuri, Kamalika and Salakhutdinov, Ruslan}, volume = {97}, series = {Proceedings of Machine Learning Research}, month = {09--15 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v97/bateni19a/bateni19a.pdf}, url = {https://proceedings.mlr.press/v97/bateni19a.html}, abstract = {In the era of big data, learning from categorical features with very large vocabularies (e.g., 28 million for the Criteo click prediction dataset) has become a practical challenge for machine learning researchers and practitioners. We design a highly-scalable vocabulary compression algorithm that seeks to maximize the mutual information between the compressed categorical feature and the target binary labels and we furthermore show that its solution is guaranteed to be within a $1-1/e \approx 63%$ factor of the global optimal solution. Although in some settings, entropy-based set functions are known to be submodular, this is not the case for the mutual information objective we consider (mutual information with respect to the target labels). To address this, we introduce a novel re-parametrization of the mutual information objective, which we prove is submodular, and also design a data structure to query the submodular function in amortized $O(\log n )$ time (where $n$ is the input vocabulary size). Our complete algorithm is shown to operate in $O(n \log n )$ time. Additionally, we design a distributed implementation in which the query data structure is decomposed across $O(k)$ machines such that each machine only requires $O(\frac n k)$ space, while still preserving the approximation guarantee and using only logarithmic rounds of computation. We also provide analysis of simple alternative heuristic compression methods to demonstrate they cannot achieve any approximation guarantee. Using the large-scale Criteo learning task, we demonstrate better performance in retaining mutual information and also verify competitive learning performance compared to other baseline methods.} }
Endnote
%0 Conference Paper %T Categorical Feature Compression via Submodular Optimization %A Mohammadhossein Bateni %A Lin Chen %A Hossein Esfandiari %A Thomas Fu %A Vahab Mirrokni %A Afshin Rostamizadeh %B Proceedings of the 36th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2019 %E Kamalika Chaudhuri %E Ruslan Salakhutdinov %F pmlr-v97-bateni19a %I PMLR %P 515--523 %U https://proceedings.mlr.press/v97/bateni19a.html %V 97 %X In the era of big data, learning from categorical features with very large vocabularies (e.g., 28 million for the Criteo click prediction dataset) has become a practical challenge for machine learning researchers and practitioners. We design a highly-scalable vocabulary compression algorithm that seeks to maximize the mutual information between the compressed categorical feature and the target binary labels and we furthermore show that its solution is guaranteed to be within a $1-1/e \approx 63%$ factor of the global optimal solution. Although in some settings, entropy-based set functions are known to be submodular, this is not the case for the mutual information objective we consider (mutual information with respect to the target labels). To address this, we introduce a novel re-parametrization of the mutual information objective, which we prove is submodular, and also design a data structure to query the submodular function in amortized $O(\log n )$ time (where $n$ is the input vocabulary size). Our complete algorithm is shown to operate in $O(n \log n )$ time. Additionally, we design a distributed implementation in which the query data structure is decomposed across $O(k)$ machines such that each machine only requires $O(\frac n k)$ space, while still preserving the approximation guarantee and using only logarithmic rounds of computation. We also provide analysis of simple alternative heuristic compression methods to demonstrate they cannot achieve any approximation guarantee. Using the large-scale Criteo learning task, we demonstrate better performance in retaining mutual information and also verify competitive learning performance compared to other baseline methods.
APA
Bateni, M., Chen, L., Esfandiari, H., Fu, T., Mirrokni, V. & Rostamizadeh, A.. (2019). Categorical Feature Compression via Submodular Optimization. Proceedings of the 36th International Conference on Machine Learning, in Proceedings of Machine Learning Research 97:515-523 Available from https://proceedings.mlr.press/v97/bateni19a.html.

Related Material