Performance of brain computer interfaces (BCI) highly depends on the how well their preprocessing and classification modules are implemented. Preprocessing allows to get cleaner signals and makes their classification easier. However, we hypothesize that information loss occurring due to the preprocessing might prevent the classifier from performing better instead of helping it. To test our hypothesis, we implemented a classifier using raw data from the 2nd Wadsworth BCI Dataset from BCI competition III and tested it with that same dataset as well as the Wadsworth BCI Dataset from BCI competition II. Furthermore, the impact in the classifier performance when augmenting the classifier training set with artificially generated data was assessed. We found that although the classifier performed similarly in most cases, the case with raw data and an augmented training dataset performed round 3 % better than the others in the same conditions on a 5-fold cross validation exercise, suggesting possible positive effects of raw and artificially generated data.

Aiming at automating the defect identification process in Akita Shizuki, a local company suffering the effects of the aging Japanese population, we present the base for a cost-efficient solution for the identification of defects on the surface of film-capacitors based on image processing and machine learning with a simple web camera. As a result of this study, a model with classification accuracy of over 90 % was achieved in the offline classification of optimal and defective film-capacitors in a 5-fold cross validation test. Testing for online classification was also performed with a limited number of physical samples, but results suggest that adjustments to our methodology need to be done to report further. Ultimately, we found that it is possible to identify defects on the surface of film capacitors with a simple offline classification system based on machine learning for the recognition of a single type of capacitor among the many produced at Akita Shizuki. Furthermore, the possibility of doing this in real time was explored, and could be pursued if there is the need for such a system in the current production line environment.

The use of electroencephalogram (EEG) data is common to develop brain-computer interface (BCI) applications. Analysis of EEG data in the oddball paradigm has revealed that some electrodes experience clearer manifestations of the P300 wave, giving a particular relevance to their position. For this study, we arrange recorded EEG data as a single trial 3D representation in which spatial and temporal information are kept with few to no modifications for off-line classification. This differs from most studies using EEG data and convolutional neural networks (CNN), for which a 2D-mapped version of the data is normally used. The recorded data used for this study corresponds to the EEG of four subjects that were presented with auditory stimuli via earphones from six different virtual directions following the oddball paradigm to elicit the P300 wave. The stimuli presentation was done at time intervals of 500, 400, and 300 ms to verify the impact of the speed in the performance of an auditory BCI. The results obtained for the identification of P300 within the EEG shows that the 3D input proposed in this study is a reliable approach for input configuration and that the CNN models trained using it tend to correctly differentiate the EEG containing the P300 from that which doesn't with low accuracy variations regardless of the differences of the used models.