Search for: All records

Motor imagery classification is known to be highly user dependent. Subspace alignment has been somewhat successful in allowing for unsupervised transfer from one training user to a new user. In this paper we develop a method to weight contributions from subspace alignment to multiple training users to give improved unsupervised transfer performance on the new test user. Ablation analyses show that both the subspace alignment and weighting are critical for improved performance. We also discuss how weighting uses the labels of the training users to better interpret subspace alignment. 

Data were collected from a brain-computer interface speller that utilized the P3b as a control signal. Stimuli consisted of letters and their “segments”. Importantly, different letters were made up of different numbers of segments from a 10 segment library. Subjects were instructed to mentally note whenever segments from their letter (targets) were flashed. We found that P3b amplitudes of target segments decreased as the number of segments in a letter (target letter complexity) increased.In contrast, the P3b attenuation was not affected by the total number of letters a segment belonged to (segment frequency).These results may reflect higher task difficulty caused by increased working memory load with increased target letter complexity. Alternatively, it’s possible that despite the target rate being fixed at 30% within each block, subjects erroneously believed the target rate increased with target letter complexity.Further work to disentangle these possibilities may enrich our understanding of the P3b. 

Brain-computer interface (BCI) systems read and infer the user brain activity directly from the brain providing a means of communication and rehabilitation for patients in need. However, brain signals are known to be non-stationary and existing systems are not reliable and robust enough to be taken outside of the laboratory. Often times long calibration and recalibration of the system is required which can be tiresome and frustrating to the user. In this study, we compare the method of common spatial patterns (CSP) with two of its variants, namely, the canonical correlation analysis approach to common spatial patterns (CCACSP) and the common spatio-spectral patterns (CSSP) in detecting the motor imagery signal when trained on calibration data with sham feedback and tested in online control. We show that the motor imagery performance is significantly better with CSSP and CCACSP compared to CSP and hence, able to provide a more reliable transfer of the classifier from calibration to online control. 

We explore the separation of decision confidence and familiarity components in EEG data from recognition memory experiments. We first develop and test a classifier designed to classify decision confidence on new trials. We then use this classifier to control for confidence in the selection of trials of familiarity and correct rejection. This allows us to reveal a familiarity component that is of similar magnitude for recollection and familiarity judgements. This familiarity component reveals more of a frontal extent than obtained without confidence matching. We believe that this preliminary result can serve as a guide for designing future electrophysiological experiments to better separate the different components of recognition memory and that the technique of using classifiers to control for response-related covariates can be used for early exploration of these components in existing data. 

Although pain is widely recognized to be a multidimensional experience, it is typically measured by unidimensional patient self-reported visual analog scale (VAS). However, self-reported pain is subjective, difficult to interpret and sometimes impossible to obtain. Machine learning models have been developed to automatically recognize pain at both the frame level and sequence (or video) level. Many methods use or learn facial action units (AUs) defined by the Facial Action Coding System (FACS) for describing facial expressions with muscle movement. In this paper, we analyze the relationship between sequence-level multidimensional pain measurements and frame-level AUs and an AU derived pain-related measure, the Prkachin and Solomon Pain Intensity (PSPI). We study methods that learn sequence-level metrics from frame-level metrics. Specifically, we explore an extended multitask learning model to predict VAS from human-labeled AUs with the help of other sequence-level pain measurements during training. This model consists of two parts: a multitask learning neural network model to predict multidimensional pain scores, and an ensemble learning model to linearly combine the multidimensional pain scores to best approximate VAS. Starting from human-labeled AUs, the model achieves a mean absolute error (MAE) on VAS of 1.73. It outperforms provided human sequence-level estimates which have an MAE of 1.76. Combining our machine learning model with the human estimates gives the best performance of MAE on VAS of 1.48. 

Brain-computer interface (BCI) systems are proposed as a means of communication for locked-in patients. One common BCI paradigm is motor imagery in which the user controls a BCI by imagining movements of different body parts. It is known that imagining different body parts results in event-related desynchronization (ERD) in various frequency bands. Existing methods such as common spatial patterns (CSP) and its refinement filterbank common spatial patterns (FB-CSP) aim at finding features that are informative for classification of the motor imagery class. Our proposed method is a temporally adaptive common spatial patterns implementation of the commonly used filter-bank common spatial patterns method using convolutional neural networks; hence it is called TA-CSPNN. With this method we aim to: (1) make the feature extraction and classification end-to-end, (2) base it on the way CSP/FBCSP extracts relevant features, and finally, (3) reduce the number of trainable parameters compared to existing deep learning methods to improve generalizability in noisy data such as EEG. More importantly, we show that this reduction in parameters does not affect performance and in fact the trained network generalizes better for data from some participants. We show our results on two datasets, one publicly available from BCI Competition IV, dataset 2a and another in-house motor imagery dataset.