A paper by Dr Rakesh Sengupta, Assistant Professor, Psychology, SIAS, titled ‘Evaluating Continuous-Time Recurrent Neural Networks for State-Dependent EEG Forecasting’ has been published in the proceedings of the 2025 2nd Asian Conference on Intelligent Technologies (ACOIT).
About the Research
The paper explores how short-term brain activity (EEG signals) can be more accurately predicted using lightweight AI models. Such forecasting is critical for real-time neurotechnologies, including Brain-Computer Interfaces (BCIs) and neurofeedback systems. The study benchmarks a Continuous-Time Recurrent Neural Network (CTRNN) against both classical methods and complex deep learning models. It finds that even a compact, highly interpretable CTRNN can effectively capture the non-linear dynamics of human brain activity, offering a competitive alternative to “black-box” AI models in real-time BCI applications.
