Engineer brings his skills to JBU projects

Vlad Ungureanu is one of the team at the Jack Birch Unit, whose bladder cancer research is supported by York Against Cancer. Here, Vlad talks about his work.

"I'm a curious engineer who recently returned to academia after two years of wandering in the industry. I graduated in 2018 from the University of York with an MEng in Electronics and Computer Engineering. I've been in the UK since 2013, during which I've been exploring different parts of the software world. My most interesting work was my Masters project under the supervision of Dr David Halliday, where I researched niche models for artificial neural networks – computer systems loosely based on the networks that make up animal brains.

"I always wanted to return to academia and do a PhD and just before the pandemic I returned to University of York to work on a few short projects with Professor Stephen Smith and the JBU (Professor Jenny Southgate and Dr Andrew Mason). One of our goals was to find patterns in the data the JBU has on benign bladder diseases as well as building a collaboration between the JBU and India. This introduced me to the world of bioinformatics and reminded me of the rigorous work that scientists do in their labs. Undoubtedly, I've enjoyed my time in the new role and I was very happy to be given the opportunity to continue being part of this environment.

"My PhD is an interdisciplinary attempt to build solutions for biomedical applications with unconventional machine learning.

"Machine learning (ML) is the way computer algorithms automatically improve through experience or by using data – it’s part of the field of artificial intelligence.

"In our project we are using evolutionary algorithms, which mimic the way nature uses mutations and natural selection to develop and improve. The project also uses spiking neural networks (SNN), which again mimic nature.

"Our motivation is to overcome the challenges of the current approaches in machine learning which are considered to be "black box" solutions. This means that it is difficult to understand how they reach an output, something that’s essential when we need to know how a gene is involved in cell proliferation or how a mutation can increase the likelihood of cancer.

"Evolutionary algorithms (EA) are considered to be a “white box” approach that makes it easier to understand the underlying connections that lead to a result.

"Traditional ML approaches also require large datasets and are resource-heavy, a limitation we may overcome with SNN.

"We are still in the early days of the project and it may seem that we have some great tools to tackle the disadvantages of the current approaches but we are at the forefront of what these new ML approaches can do. They come with many challenges which we need to address.

"This project is an exciting one to work on and I am fortunate to be supervised by leading experts in their field: Prof Southgate and Dr Mason in bladder cancer and bioinformatics, and Prof Smith and Dr Halliday on the computational side, evolutionary algorithms and spiking neural networks respectively."