Coloquio: Ciencia de datos y machine learning para la caracterización de planetas extrasolares
- 2024-08-15 10:30 |
- Aula Federman
Jueves 21/5/2009, 14 hs
Aula Federman, 1er piso, Pabellón I
During normal function neurons in the brain produce ionic currents that can be measured as electric field potentials. These signals provide an unique window into the temporal aspects of neuronal activity. They allow us to monitor, and indeed, modulate neuronal timing. Since we are interested in the human brain we use non-invasive electrodes and have to cope with low signal-to-noise ratios (SNR). The talk will present a framework for extracting single-trial information from functional imaging data with low SNR. When applied to surface-electrode signals (EEG) these tools are remarkably successful at extracting timing information, surpassing in fact the temporal accuracy of human behavioral response. We have given a dramatic demonstration of this by using these timing signals for a brain-computer interface in the context of broad-area image search. For this task we were able to show 400% improvement in human search performance.