Deep Kalman Filters

TL;DR

This paper introduces a unified deep learning approach to Kalman filters, enabling efficient learning and counterfactual inference in time-series data, demonstrated on synthetic and real-world medical datasets.

Contribution

It presents a novel variational method for learning deep Kalman filters, extending their application to counterfactual inference in complex temporal data.

Findings

Effective modeling of the Healing MNIST dataset.

Successful counterfactual inference on electronic health records.

Demonstrated scalability to large patient datasets.

Abstract

Kalman Filters are one of the most influential models of time-varying phenomena. They admit an intuitive probabilistic interpretation, have a simple functional form, and enjoy widespread adoption in a variety of disciplines. Motivated by recent variational methods for learning deep generative models, we introduce a unified algorithm to efficiently learn a broad spectrum of Kalman filters. Of particular interest is the use of temporal generative models for counterfactual inference. We investigate the efficacy of such models for counterfactual inference, and to that end we introduce the "Healing MNIST" dataset where long-term structure, noise and actions are applied to sequences of digits. We show the efficacy of our method for modeling this dataset. We further show how our model can be used for counterfactual inference for patients, based on electronic health record data of 8,000 patients over 4.5 years.