Discriminative Regularization for Latent Variable Models with Applications to Electrocardiography

Andrew Miller, Ziad Obermeyer, John Cunningham, Sendhil Mullainathan
Proceedings of the 36th International Conference on Machine Learning, PMLR 97:4585-4594, 2019.

Abstract

Generative models often use latent variables to represent structured variation in high-dimensional data, such as images and medical waveforms. However, these latent variables may ignore subtle, yet meaningful features in the data. Some features may predict an outcome of interest (e.g. heart attack) but account for only a small fraction of variation in the data. We propose a generative model training objective that uses a black-box discriminative model as a regularizer to learn representations that preserve this predictive variation. With these discriminatively regularized latent variable models, we visualize and measure variation in the data that influence a black-box predictive model, enabling an expert to better understand each prediction. With this technique, we study models that use electrocardiograms to predict outcomes of clinical interest. We measure our approach on synthetic and real data with statistical summaries and an experiment carried out by a physician.

Cite this Paper

BibTeX
@InProceedings{pmlr-v97-miller19a, title = {Discriminative Regularization for Latent Variable Models with Applications to Electrocardiography}, author = {Miller, Andrew and Obermeyer, Ziad and Cunningham, John and Mullainathan, Sendhil}, booktitle = {Proceedings of the 36th International Conference on Machine Learning}, pages = {4585--4594}, 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/miller19a/miller19a.pdf}, url = {https://proceedings.mlr.press/v97/miller19a.html}, abstract = {Generative models often use latent variables to represent structured variation in high-dimensional data, such as images and medical waveforms. However, these latent variables may ignore subtle, yet meaningful features in the data. Some features may predict an outcome of interest (e.g. heart attack) but account for only a small fraction of variation in the data. We propose a generative model training objective that uses a black-box discriminative model as a regularizer to learn representations that preserve this predictive variation. With these discriminatively regularized latent variable models, we visualize and measure variation in the data that influence a black-box predictive model, enabling an expert to better understand each prediction. With this technique, we study models that use electrocardiograms to predict outcomes of clinical interest. We measure our approach on synthetic and real data with statistical summaries and an experiment carried out by a physician.} }
Endnote
%0 Conference Paper %T Discriminative Regularization for Latent Variable Models with Applications to Electrocardiography %A Andrew Miller %A Ziad Obermeyer %A John Cunningham %A Sendhil Mullainathan %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-miller19a %I PMLR %P 4585--4594 %U https://proceedings.mlr.press/v97/miller19a.html %V 97 %X Generative models often use latent variables to represent structured variation in high-dimensional data, such as images and medical waveforms. However, these latent variables may ignore subtle, yet meaningful features in the data. Some features may predict an outcome of interest (e.g. heart attack) but account for only a small fraction of variation in the data. We propose a generative model training objective that uses a black-box discriminative model as a regularizer to learn representations that preserve this predictive variation. With these discriminatively regularized latent variable models, we visualize and measure variation in the data that influence a black-box predictive model, enabling an expert to better understand each prediction. With this technique, we study models that use electrocardiograms to predict outcomes of clinical interest. We measure our approach on synthetic and real data with statistical summaries and an experiment carried out by a physician.
APA
Miller, A., Obermeyer, Z., Cunningham, J. & Mullainathan, S.. (2019). Discriminative Regularization for Latent Variable Models with Applications to Electrocardiography. Proceedings of the 36th International Conference on Machine Learning, in Proceedings of Machine Learning Research 97:4585-4594 Available from https://proceedings.mlr.press/v97/miller19a.html.

Related Material