Dataset Information

A Comparison of Regularization Methods in Forward and Backward Models for Auditory Attention Decoding.

ABSTRACT: The decoding of selective auditory attention from noninvasive electroencephalogram (EEG) data is of interest in brain computer interface and auditory perception research. The current state-of-the-art approaches for decoding the attentional selection of listeners are based on linear mappings between features of sound streams and EEG responses (forward model), or vice versa (backward model). It has been shown that when the envelope of attended speech and EEG responses are used to derive such mapping functions, the model estimates can be used to discriminate between attended and unattended talkers. However, the predictive/reconstructive performance of the models is dependent on how the model parameters are estimated. There exist a number of model estimation methods that have been published, along with a variety of datasets. It is currently unclear if any of these methods perform better than others, as they have not yet been compared side by side on a single standardized dataset in a controlled fashion. Here, we present a comparative study of the ability of different estimation methods to classify attended speakers from multi-channel EEG data. The performance of the model estimation methods is evaluated using different performance metrics on a set of labeled EEG data from 18 subjects listening to mixtures of two speech streams. We find that when forward models predict the EEG from the attended audio, regularized models do not improve regression or classification accuracies. When backward models decode the attended speech from the EEG, regularization provides higher regression and classification accuracies.

SUBMITTER: Wong DDE

PROVIDER: S-EPMC6090837 | biostudies-literature | 2018

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

A Comparison of Regularization Methods in Forward and Backward Models for Auditory Attention Decoding.

Wong Daniel D E DDE Fuglsang Søren A SA Hjortkjær Jens J Ceolini Enea E Slaney Malcolm M de Cheveigné Alain A

Frontiers in neuroscience 20180807

The decoding of selective auditory attention from noninvasive electroencephalogram (EEG) data is of interest in brain computer interface and auditory perception research. The current state-of-the-art approaches for decoding the attentional selection of listeners are based on linear mappings between features of sound streams and EEG responses (forward model), or vice versa (backward model). It has been shown that when the envelope of attended speech and EEG responses are used to derive such mappi ...[more]

PMID: 30131670

Similar Datasets

Project description:ImportanceThe association of fast backward-rotating shift work (ie, anticlockwise sequence of afternoon, morning, and night shifts) with subjective and objective measures of sleep-wake quality, daytime attention, and tiredness of health care workers has not yet been established.ObjectiveTo investigate the association of shift rotation direction with tiredness, sleepiness, and sustained attention among nurses working forward- and backward-rotating shifts.Design, setting, and participantsData of this cohort study were collected from nurses working at 5 midsized Italian hospitals. The nurses had either a forward-rotating schedule (ie, morning to afternoon to night) and or a backward-rotating schedule (ie, afternoon to morning to night). The data were collected from July 2017 to February 2020. Data analysis was performed from May to October 2020.ExposuresParticipants were working either forward- or backward-rotating schedules, in which the sequence of 3 shifts (morning, afternoon, and night) changed in a clockwise or anticlockwise direction.Main outcomes and measuresSleep data were collected using the Karolinska Sleepiness Scale and Pittsburgh Sleep Quality Index. Sustained attention was measured using the Psychomotor Vigilance Task. Tiredness was evaluated using the Tiredness Symptom Scale.ResultsA total of 144 nurses (mean [SE] age, 41.3 [0.8] years; 92 women [63.9%]) participated in the study; 80 nurses had forward-rotating schedules, and 64 had backward-rotating schedules. Nurses with irregular sleep-wake patterns due to night shift work had poor sleep quality (46 [57.5%] in forward-rotating schedule group; 37 [57.8%] in backward-rotating schedule group). Nurses working backward-rotating shifts exhibited significantly greater sleepiness (F1,139 = 41.23; P < .001) and cognitive slowing (ie, longer median reaction times; F1,139 = 42.12; P < .001) than those working forward rotations. Importantly, these differences were not affected by age, years of employment, and quality of sleep. Of nurses working on a backward-rotating schedule, 60 (93.8%) reported elevated sleepiness (Karolinska Sleepiness Scale score ≥7) after the night shift. The median reaction time (F1,139 = 42.12; P < .001), 10% fastest reaction time (F1,139 = 97.07; P < .001), minor lapses (F1,139 = 46.29; P < .001), and reaction time distribution (F1,139 = 60.13; P < .001) of nurses on backward-rotating schedules indicated a lower level of vigilance, which is negatively associated with neurobehavioral performance.Conclusions and relevanceIn this study, both shift rotation models were negatively associated with health and cognitive performance. These findings suggest that forward shift rotation may be more beneficial than backward rotation for several measured performance attentional outcomes and sleepiness. Optimization of shift rotations should be implemented to decrease the combination of the negative outcomes associated with shift work and reduce the potential risk of medical errors in health care systems.

Dataset Information

A Comparison of Regularization Methods in Forward and Backward Models for Auditory Attention Decoding.

Publications

A Comparison of Regularization Methods in Forward and Backward Models for Auditory Attention Decoding.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets