PhD Profile: Hamzeh Asgharnezhad

What led you to this topic?

What led me to this topic was the realization that deep learning models often make confident predictions even when they’re uncertain—which can be dangerous in critical areas like healthcare. I wanted to explore how we can make these models more responsible and reliable by teaching them to abstain from making decisions when they’re unsure. This gap in current AI systems sparked my interest and led me to focus on uncertainty-aware deep learning.

Why is your research important and how will it make a difference?

My research is important because it addresses a critical vulnerability in current deep learning systems: their inability to handle uncertainty responsibly, especially in high-risk, real-world applications such as medical diagnosis, autonomous systems, and cybersecurity.

What inspired you to start a career in research?

I started a career in research because I’ve always had a passion for learning and discovering new things. In my previous work, I felt something was missing – I was applying existing knowledge but not creating anything new. I wanted more. Research gives me the chance to go deeper, ask meaningful questions, and work on innovative solutions that can make a real impact.

Why did you choose to complete a PhD at IISRI?

I chose to pursue a PhD at IISRI because of its strong emphasis on practical, real-world applications of research. What stood out to me is that IISRI doesn’t just focus on theoretical development—it actively works on solving real problems with real impact, especially in areas like medical diagnosis and instruments, and intelligent systems.

What have been your highlights so far?

Teamwork and collaboration – I successfully worked in a multidisciplinary research team, contributing to shared goals while learning from peers with diverse technical backgrounds. Also understanding new models and approaches by gaining deep insights into modern deep learning architectures, uncertainty modeling, and abstention mechanisms—enhancing my ability to critically evaluate and develop novel solutions. And exposure to real-world problems. I collaborated with the Department of Health, which gave me valuable perspective on real-world data challenges, especially in medical and healthcare domains, and how AI solutions must be designed with safety, ethics, and impact in mind

What would you like to do once your PhD is completed?

My goal is to be part of a team—whether in academia, industry, or a research lab—that focuses on developing trustworthy AI systems for high-stakes applications like healthcare, safety, or intelligent decision-making. I’m especially passionate about addressing the real-world limitations of current models, such as their inability to manage uncertainty, and using my research to build systems that are not only smart but also safe, and reliable.

This article was originally published in the 2024 IISRI Annual Report.