PhD candidate Stefan Schliebs has spent the past three years investigating how the human brain works and whether it's possible to create machines capable of performing the same function.
Understanding how the human brain works and whether it's possible to create machines capable of performing the same functions has occupied PhD candidate Stefan Schliebs' brain for the past three years.
At AUT's Knowledge Engineering Discovery Research Institute (KEDRI), Schliebs and other researchers have been looking for answers by replicating the human brain within the structure of a computer. Our brains are highly complex and dense networks consisting of approximately 100 billion interconnected elementary processing units called neurons. Neurons communicate with each other through the exchange of short electrical pulses known as spikes.
"It's been found that the brain's neural networks can be imitated in the medium of silicon and electricity that makes up a computer and substantial progress has been made in areas such as speech recognition, learning rules and associative memory."
Many studies are now attempting to use neural networks for practical applications to solve complex real world problems. In his recently completed PhD thesis Heterogneous Probabilistic Models for Optimisation and Modelling of Evolving Neural Networks, Schliebs looks at a new neural networking model (evolving spiking neural networks which were developed at KEDRI) and builds new algorithms to optimise its function resulting in speedier and more accurate results.
He worked with Lincoln University and the Centre for Bioprotection and applied his new method to predict how a certain species of fruit fly could be expected to invade New Zealand. His model, which optimises data features and changes its parameters, is capable of significantly increasing the prediction accuracy compared to current approaches. Now with his thesis under review and the publication of eight articles in journals and conferences under his belt, Schliebs is considering his next move.
A native of Germany, where he did a masters degree in computer science at the University of Leipzig, he came to New Zealand at the invitation of KEDRI director Professor Nikola Kasabov who was looking for a person to research quantum inspired algorithms.
"When I came Professor Kasabov had some ideas but gave me a lot of space to grow. It was very important to me to have freedom and not a strictly defined plan. I also learned a lot from looking over the shoulder of my second supervisor Michael Defoin-Platel. We fitted together very well."
With the possibility of moving to industry or continuing academic work, Schliebs says he is drawn to the diversity of the academic lifestyle. Schliebs still has more questions he wants answered, too, which he hopes will bring the world closer to its own answers by combining the processing powers of computers with the brain's ability to remember, learn, sense and communicate.
Flys, brains and artificial intelligence - a quantum puzzle for some... Click here to find out more.