Foundations of Pattern Recognition & Trustworthy Machine Learning

I study the mathematical and computational foundations of pattern recognition and machine learning: how learning systems extract, preserve, and integrate information from complex data. My work addresses representation learning, information theory, feature complementarity, and the robustness and interpretability of learned models. Cardiovascular health is the principal application domain, where these questions are studied with ECG and photoplethysmography (PPG) signals.

Research Publications Lab
Portrait of Dr. Timothy Oladunni

Flagship Research Project

Latest News

NIH Awards Morgan State $520,000 Grant to Advance Biomedical Artificial Intelligence Research

October 29, 2026 · Morgan State University News

Dr. Oladunni is Principal Investigator of ORBIL-AI, a four-year NIH Support for Research Excellence (SuRE) R16 award administered by the National Library of Medicine. Read the announcement → · Project page →

Research Focus

Foundations

Representation & Information

How learned representations capture, preserve, and combine information. Topics include feature complementarity and redundancy across signal domains, information-theoretic measures of multimodal fusion, and dynamical-systems descriptions of physiological signals.

Foundations

Trustworthy Machine Learning

When a model's prediction should and should not change. Topics include robustness to benign physiological variability, concept drift, the faithfulness of explanations, and evaluation practice that avoids inflated performance estimates.

Application

Cardiovascular Health

ECG and PPG analysis, cardiovascular stability assessment, cuffless blood pressure estimation, and performance across skin tones and demographic groups. Clinical and wearable-device validation are ongoing; see the Research page for the status of each line of work.

At a Glance

12

Peer-reviewed journal articles

13

Peer-reviewed conference papers

8

Preprints (2025–2026)

4

Manuscripts under review

9

Theoretical research frameworks

Counts reflect the Publications page. The 4 manuscripts under review also have public preprints on arXiv but are counted separately from the 8 preprints. Neither preprints nor manuscripts under review have completed peer review.

Research Overview

Research Vision

My research investigates fundamental questions in representation learning, information theory, dynamical systems, and trustworthy machine learning through nine distinct but related theoretical frameworks. Each framework addresses a specific scientific problem, from information complementarity and explanation reliability to physiological invariance, dynamical stability, and longitudinal monitoring. Together, these directions establish a research program connecting foundational computer science with biomedical information analysis.

Several frameworks are published in peer-reviewed venues; others are available as preprints or remain under investigation. The Research page gives the status of each.

Teaching Philosophy

My teaching is grounded in mathematical reasoning and the foundations of computer science. I connect theory to computation through hands-on experimentation, so that students learn to analyze problems, test their ideas, and solve problems independently. Research-based learning is central: students work on open questions, design and evaluate computational experiments, and communicate their findings. Through close mentorship, I prepare students for graduate study and advanced research.