��ݮ����Ƶ����

Taylor Kergan is helping make history at the at the ��ݮ����Ƶ����. He’s the first graduate of the institute’s new Computational Neuroscience (CN) interdisciplinary specialization program. Launched in fall 2022, the program opened new areas of exploration and learning for Kergan, who combined the specialization with his pursuit of graduate studies in physics.

Drawn to computational neuroscience through its fascinating coursework, Kergan’s research took a deep dive into brain-inspired machine learning architectures. Machine learning is a type of technology that allows computers to learn from data and improve over time without being directly programmed. Kergan’s thesis explored optimizing neural networks using evolutionary algorithms and particle swarm optimization, aiming to initialize networks in their best possible state. His long-term goal? To use machine-learned models for diagnosis and exploration in brain research.

“We created the Computational Neuroscience specialization to fill a need we saw for transdisciplinary training and to bring together faculty and students in physics, neuroscience, psychology and computer science,” says PhD, who helped develop the program, 

“This field allows us to better understand the brain, while also developing new technologies. By training students to think across these disciplines, we are opening the door to transformative insights and innovations.”

One of Kergan’s proudest achievements was studying the "edge of chaos" — a point where systems like the brain are thought to operate for maximum flexibility and efficiency. Imagine it as the balance between complete randomness and perfect order. Kergan’s work showed that machine learning networks operate most effectively just below this chaotic point. 

“In simpler terms, the network’s ability to learn and predict patterns, like identifying the next point on a graph, was strongest when the connections in the network were slightly more stable than what is expected,” says Kergan. “This finding has important implications for creating more efficient and powerful AI models.”

Kergan’s supervisor, believes these findings could also lead to more precise models for understanding neural behaviour. 

“This research truly highlights the power of computational neuroscience,” says Nicola, “Taylor’s work offers valuable insights into how the brain itself functions. By mimicking the brain’s processes in a controlled way, this research lays the groundwork toward future models that will be able to predict outcomes in brain health by using a model neural network.”

Currently pursuing a PhD in electrical and computer engineering at the University of California, Santa Cruz, Kergan is expanding his research into brain-inspired hardware. He is passionate about creating AI systems that use less energy. AI currently requires massive amounts of electricity to power the complex computations that drive it, leading to a significant carbon footprint. Kergan’s PhD research aims to develop hardware that produces the same — if not better — results while using far less energy. This could revolutionize the tech industry by making AI greener and more sustainable.

Taylor encourages others to explore computational neuroscience, whether from a neuro or non-neuro background, for its unique intersection of biology, data, and human impact. 

“The comp-neuro program allowed me to approach my research with a more well-rounded background,” he says, “My overall knowledge base was expanded by adding comp-neuro, pushing me outside my comfort zone and I am better for it!”