We use cookies to ensure that we give you the best experience on our website. You can change your cookie settings at any time. Otherwise, we'll assume you're OK to continue.

Durham University

Computer Science


Publication details for Professor Toby Breckon

Podmore, J.J., Breckon, T.P., Aznan, N.K.N. & Connolly, J.D. (2019). On the Relative Contribution of Deep Convolutional Neural Networks for SSVEP-based Bio-Signal Decoding in BCI Speller Applications. IEEE Transactions on Neural Systems & Rehabilitation Engineering 27(4): 611-618.

Author(s) from Durham


Brain-computer interfaces (BCI) harnessing Steady State Visual Evoked Potentials (SSVEP) manipulate the frequency and phase of visual stimuli to generate predictable oscillations in neural activity. For BCI spellers, oscillations are matched with alphanumeric characters allowing users to select target numbers and letters. Advances in BCI spellers can, in part, be accredited to subject-speci?c optimization, including; 1) custom electrode arrangements, 2) ?lter sub-band assessments and 3) stimulus parameter tuning. Here we apply deep convolutional neural networks (DCNN) demonstrating cross-subject functionality for the classi?cation of frequency and phase encoded SSVEP. Electroencephalogram (EEG) data are collected and classi?ed using the same parameters across subjects. Subjects ?xate forty randomly cued ?ickering characters (5 ×8 keyboard array) during concurrent wet-EEG acquisition. These data are provided by an open source SSVEP dataset. Our proposed DCNN, PodNet, achieves 86% and 77% of?ine Accuracy of Classi?cation across-subjects for two data capture periods, respectively, 6-seconds (information transfer rate= 40bpm) and 2-seconds (information transfer rate= 101bpm). Subjects demonstrating sub-optimal (< 70%) performance are classi?ed to similar levels after a short subject-speci?c training period. PodNet outperforms ?lter-bank canonical correlation analysis (FBCCA) for a low volume (3channel) clinically feasible occipital electrode con?guration. The networks de?ned in this study achieve functional performance for the largest number of SSVEP classes decoded via DCNN to date. Our results demonstrate PodNet achieves cross-subject, calibrationless classi?cation and adaptability to sub-optimal subject data and low-volume EEG electrode arrangements.