Publications

Ventricular tachycardia (VT) is a major cause of sudden cardiac deaths, treated with catheter ablation that targets and destroys the tissue responsible for initiating or sustaining VT. Pace-mapping is a key practice for identifying ablation targets by matching stimulated sites’ ECG to the observed clinical VT. An AI-assisted pace-mapping approach can help clinicians identify the most likely target sites, reducing procedure time and increasing the success rate.

Deep active learning (DAL) studies the optimal selection of labeled data for training deep neural networks (DNNs). While data selection in traditional active learning is mostly optimized for given features, in DNN these features are learned and change with the learning process as well as the choices of DNN architectures. How is the optimal selection of data affected by this change is not well understood in DAL.

High-performance scientific simulations, important for comprehension of complex systems, encounter computational challenges especially when exploring extensive parameter spaces. There has been an increasing interest in developing deep neural networks (DNNs) as surrogate models capable of accelerating the simulations. However, existing approaches for training these DNN surrogates rely on extensive simulation data which are heuristically selected and generated with expensive computation – a challenge under-explored in the literature.

Personalized cardiac models are crucial intervention tools for a multitude of cardiac health issues. As cardiac simulations become more complex and expensive, machine learning (ML) models demonstrated the potential to enable efficient model personalization and cardiac tissue parameter estimation.

We present a Bayesian active learning method to directly approximate the posterior pdf function of cardiac model parameters, in which we intelligently select training points to query the simulation model in order to learn the posterior pdf using a small number of samples. We integrate a generative model into Bayesian active learning to allow approximating posterior pdf of high-dimensional model parameters at the resolution of the cardiac mesh.

In this work, we argue that regularizing the global smoothness of neural functions by filling the void in between data points can further improve SSL. We present a novel SSL approach that trains the neural network on linear mixing of labeled and unlabeled data, at both the input and latent space in order to regularize different portions of the network.

We present a novel concept that uses a generative variational auto-encoder (VAE) to embed HD Bayesian optimization into a low-dimensional (LD) latent space that represents the generative code of HD parameters. We further utilize VAE-encoded knowledge about the generative code to guide the exploration of the search space.

Indoor localization and mapping is an important problem with many applications such as emergency response, architectural modeling, and historical preservation. In this project, a metrically accurate, GPS-denied, indoor 3D static mapping system was developed using a moveable base coupled with three degree of freedom IMU.