Electronic device for recognition of mental behavioral attributes based on deep neural networks

An electronic device, comprising:

• a memory configured to store a set of electroencephalogram (EEG) signals, a set of bio-signals, and a plurality of user parameters associated with a specific subject;
• and
• at least one processor configured to:
• input the set of EEG signals and the set of bio-signals associated with the specific subject to a plurality of first recognition models pre-trained on a training set of EEG signals and a training set of different types of bio-signals associated with a set of training subjects, wherein
• the plurality of first recognition models includes a first plurality of deep neural networks (DNNs),
• a first DNN of the plurality of DNNs is pre-trained based on a first training set of EEG signals of the training set of EEG signals,
• each EEG signal of the first training set of EEG signals corresponds to a first region of a brain of a respective training subject of the plurality of training subjects,
• a second DNN of the plurality of DNNs is pre-trained based on a second training set of EEG signals of the training set of EEG signals,
• each EEG signal of the second training set of EEG signals corresponds to a second region of the brain of the respective training subject of the plurality of training subjects,
• the first region is different from the second region, and
• the set of training subjects is different from the specific subject;
• estimate a plurality of relationship data, which indicates one of a functional connectivity or a dependency between a first region of a brain of the specific subject and a second region of the brain of the specific subject,
• wherein the plurality of relationship data is estimated by application of the pre-trained plurality of first recognition models on the set of EEG signals associated with the specific subject and the set of bio-signals associated with the specific subject;
• and
• identify a specific mental behavioral attribute of the specific subject from a set of mental behavioral attributes, by application of an output layer of a second recognition model on the estimated plurality of relationship data, wherein
• the second recognition model is pre-trained on a feature vector from output layers of the plurality of first recognition models,
• the second recognition model is pre-trained based on an interconnectivity parameter and a dependency parameter between a first set of mental behavioral attributes of the set of mental behavioral attributes and a second set of mental behavioral attributes of the set of mental behavioral attributes,
• the interconnectivity parameter indicates the functional connectivity between the first region of the brain of the specific subject and the second region of the brain of the specific subject in a case in which the brain of the specific subject stays functional for at least two mental behavioral attributes of the set of mental behavioral attributes,
• the first region of the brain of the specific subject is different from the second region of the brain of the specific subject,
• the first region of the brain of the specific subject corresponds to an emotional state of the specific subject,
• the second region of the brain of the specific subject corresponds to a body movement of the specific subject, and
• the dependency parameter indicates an effect of a first mental behavioral attribute of the set of mental behavioral attributes on a second mental behavioral attribute of the set of mental behavioral attributes.

A method, comprising:

• in an electronic device that comprises a memory and at least one processor:
• storing, by the memory, a set of electroencephalogram (EEG) signals, a set of bio-signals, and a plurality of user parameters associated with a specific subject;
• training, by the at least one processor, a plurality of first recognition models, that includes a first plurality of deep neural networks (DNNs), on a training set of EEG signals and a training set of bio-signals associated with a set of training subjects, wherein
• a first DNN of the plurality of DNNs is trained based on a first training set of EEG signals of the training set of EEG signals,
• each EEG signal of the first training set of EEG signals corresponds to a first region of a brain of a respective training subject of the plurality of training subjects,
• a second DNN of the plurality of DNNs is trained based on a second training set of EEG signals of the training set of EEG signals,
• each EEG signal of the second training set of EEG signals corresponds to a second region of the brain of the respective training subject of the plurality of training subjects,
• the first region is different from the second region, and
• the set of training subjects is different from the specific subject;
• training, by the at least one processor, a second recognition model that corresponds to a second plurality of deep neural networks (DNNs), on a feature vector from output layers of the plurality of first recognition models, wherein
• the second recognition model is trained based on an interconnectivity parameter and a dependency parameter between a first set of mental behavioral attributes of a set of mental behavioral attributes and a second set of mental behavioral attributes of the set of mental behavioral attributes,
• the interconnectivity parameter indicates a functional connectivity between a first region of a brain of the specific subject and a second region of the brain of the specific subject in a case in which the brain of the specific subject stays functional for at least two mental behavioral attributes of the set of mental behavioral attributes,
• the first region of the brain of the specific subject is different from the second region of the brain of the specific subject,
• the first region of the brain of the specific subject corresponds to an emotional state of the specific subject,
• the second region of the brain of the specific subject corresponds to a body movement of the specific subject, and
• the dependency parameter indicates an effect of a first mental behavioral attribute of the set of mental behavioral attributes on a second mental behavioral attribute of the set of mental behavioral attributes;
• estimating, by the at least one processor, a plurality of relationship data, which indicates one of the functional connectivity or a dependency between the first region of the brain of the specific subject and the second region of the brain of the specific subject,
• wherein the plurality of relationship data is estimated by application of the trained plurality of first recognition models on the set of EEG signals associated with the specific subject and the set of bio-signals associated with the specific subject;
• and
• identifying, by the at least one processor, a specific mental behavioral attribute of the specific subject from the set of mental behavioral attributes, by application of an output layer of the trained second recognition model on the estimated plurality of relationship data.