We congratulate Dr.Uğur Teğin and his student Bahadır Utku Kesgin for their study

Electrical and Electronics Engineering faculty member Dr. Uğur Teğin and his student Bahadır Utku Kesgin have created a new analog computing method that uses chaotic nonlinear attractors for efficient machine learning. Their study, published in Communication Engineering of Nature Portfolio and featured in Nature's "Neuromorphic Hardware and Computing 2024" collection, introduces a low-power, highly accurate analog model inspired by neuromorphic computing. 

New computing techniques are being developed to reduce the energy demand of modern artificial intelligence. In this study, the authors proposed chaotic and nonlinear systems to replace current machine learning algorithms. They discovered that Lorenz Chaos, famous for the Butterfly effect, is ideal for such an application. This model performs fundamental machine learning tasks at milliwatt-scale power levels while matching the performance of current techniques. By leveraging the unique properties of chaotic attractors, this work offers a sustainable and efficient solution to the power consumption issues in machine learning and marks a significant advance in analog computing. Read more https://rdcu.be/dNSeM