Portfolio item number 1
Short description of portfolio item number 1
Posts by Collection
portfolio
Portfolio item number 2
Short description of portfolio item number 2
publications
Investigating the Predictive Reproducibility of Federated Graph Neural Networks Using Medical Datasets
Published in Predictive Intelligence in Medicine: 5th International Workshop, PRIME 2022, Held in Conjunction with MICCAI, 2022
In this work, we investigate the predictive reproducibility of federated graph neural networks (GNNs) using medical datasets. We evaluate the performance of various GNN models in a federated learning setting and analyze their performance.
Interpretable Deep Learning for Myocardial Infarction Detection from ECG Signals
Published in 31st Signal Processing and Communications Applications Conference (SIU), 2023
In this study, we show that deep learning models can be trained to detect myocardial infarction (MI) from 12-lead ECG signals. We also show that the model can be made interpretable by using gradient class activation maps (Grad-CAMs) to highlight the segments of the ECG that contribute most to the decision.
Interpretable ECG analysis for myocardial infarction detection through counterfactuals
Published in Biomedical Signal Processing and Control, 2025
In this study, we propose a novel method for utilizing counterfactual explanations in ECG analysis, specifically for the detection of myocardial infarction. Our approach leverages the PTB-XL dataset and incorporates systematic feature extraction and refinement techniques to enhance interpretability for clinicians. The Visualizing Counterfactual Clues on Electrocardiograms (VCCE) method aims to bridge the gap between advanced data analysis and clinical decision-making.
Bayesian Basis Function Approximation for Scalable Gaussian Process Priors in Deep Generative Models
Published in Proceedings of the 42nd International Conference on Machine Learning (ICML), 2025
We propose a new way of utlizing Gaussian process priors in deep generative models, specifically variational autoencoders, with global parameterization that avoids explicit kernels, runs in linear time, eliminates the amortization gap, overcomes the limitations of categorical inducing point optimization, enhances interpretability in the latent space by quantifying the contributions of different covariates/effects using Sobol indices, allows for standard mini‑batch training, and treats kernel hyperparameters probabilistically.
talks
Talk 1 on Relevant Topic in Your Field
Published:
This is a description of your talk, which is a markdown file that can be all markdown-ified like any other post. Yay markdown!
Conference Proceeding talk 3 on Relevant Topic in Your Field
Published:
This is a description of your conference proceedings talk, note the different field in type. You can put anything in this field.
teaching
Teaching experience 1
Undergraduate course, University 1, Department, 2014
This is a description of a teaching experience. You can use markdown like any other post.
Teaching experience 2
Workshop, University 1, Department, 2015
This is a description of a teaching experience. You can use markdown like any other post.