Rate My Professor Lech Szymanski

Szymanski's research focuses on machine learning, deep representation and connectionist models, convolutional neural networks, reinforcement learning, support vector machines, and learning theory. He develops autonomous machine learning algorithms that form appropriate models independently, without requiring expert specification of architecture, parameters, or representations. Key publications include "Pseudo-rehearsal: Achieving deep reinforcement learning without catastrophic forgetting" (Atkinson, McCane, Szymanski, and Robins, Neurocomputing, 2021), "Deep Networks are Effective Encoders of Periodicity" (Szymanski and McCane, IEEE Transactions on Neural Networks and Learning Systems, 2014), "Sequential Learning in the Dense Associative Memory" (McAlister, Robins, and Szymanski, Neural Computation, 2025), "Prototype Analysis in Hopfield Networks with Hebbian Learning" (McAlister, Robins, and Szymanski, Neural Computation, 2024), "Conceptual complexity of neural networks" (Szymanski, McCane, and Atkinson, Neurocomputing, 2022), and "Image Segmentation with a Deep Declarative Network" (Szymanski, Mills, Wales, and McAlister, Proceedings of IVCNZ, 2024). He teaches courses including COSC343 Artificial Intelligence, COSC420 Neural Networks, and COSC360 Computer Game Design, and is a member of the Otago AI Group.