Brain-based thought identifier and classifier

Patent metadata

Field	Value
Publication number	US-2019336024-A1
Application number	US-201815972312-A
Country	US
Kind code	A1
Filing date	May 7, 2018
Priority date	May 7, 2018
Publication date	Nov 7, 2019
Grant date	—

Abstract

Embodiments provide thought classifier devices that convert analog electroencephalogram signals obtained from mental activity of a person to a digital signal bitstream data; identify a portion of the bitstream as a thought chunk representing discrete thought activity in response to matching, via a first artificial neural network comparison, digital signal bitstream thought chunk portion metadata to metadata labeled in association with a thought within a thoughts data set, the first artificial neural network trained on the thoughts data set; identify a user category in response to matching, via a different, second artificial neural network comparison, metadata of the thought chunk portion to labeled metadata within the thoughts data set, the second artificial neural network trained on the thoughts data set; and identify a specific thought of the thoughts data that has metadata that has corresponding metadata.

First claim

What is claimed is:

1 . A computer-implemented method for a thought classifier, the method comprising executing on a computer processor:
converting analog electroencephalogram signals to a digital signal bitstream data, wherein the analog electroencephalogram signals are obtained from mental activity of a person; identifying a portion of the digital signal bitstream data as a thought chunk that represents discrete thought activity in response to matching, via a first artificial neural network comparison, first metadata of the thought chunk portion of the digital signal bitstream to metadata that is labeled in association with a thought within a thoughts data set, wherein the first artificial neural network is trained on the thoughts data set; identifying a user category to which the person generating the mental activity belongs in response to matching, via a second artificial neural network comparison, the first metadata and length and generation time metadata of the thought chunk portion to metadata that is labeled in association with a user category within the thoughts data set, wherein the second artificial neural network is trained on the thoughts data set and is different from the first artificial neural network; and identifying a specific thought within the thoughts data that has metadata that corresponds to the first metadata, the length and generation time metadata of the thought chunk portion, and to metadata that is labeled in association with the identified user category within the thoughts data set.

2 . The method of claim 1 , further comprising:
integrating computer-readable program code into a computer system comprising a processor, a computer readable memory in circuit communication with the processor, and a computer readable storage medium in circuit communication with the processor; and wherein the processor executes program code instructions stored on the computer-readable storage medium via the computer readable memory and thereby performs the converting the analog electroencephalogram signals to the digital signal bitstream data, the identifying the portion of the digital signal bitstream data as the thought chunk in response to the matching via the first artificial neural network comparison, the identifying the user category in response to the matching via the second artificial neural network comparison, and the identifying the specific thought within the thoughts.

3 . The method of claim 2 , wherein the computer-readable program code is provided as a service in a cloud environment.

4 . The method of claim 1 , wherein the identifying the specific thought comprises:
matching, via a third artificial neural network comparison, the first metadata, the length and generation time metadata of the thought chunk portion and the metadata that is labeled in association with the identified user category to metadata that is labeled in association with the specific thought within the thoughts data set, wherein the third artificial neural network is trained on the thoughts data set and is different from the first artificial neural network and from the second artificial neural network.

5 . The method of claim 4 , wherein the first metadata of the thought chunk portion comprises metadata of the analog electroencephalogram signals that are converted into the thought chunk portion of the digital signal bitstream that is selected from the group consisting of frequency values of the analog electroencephalogram signals, fast-Fourier transform values of the frequency values, amplitude values of the analog electroencephalogram signals, fast-Fourier transform values of the amplitude values, voltage values of the analog electroencephalogram signals, and signal length values of the analog electroencephalogram signals.

6 . The method of claim 5 , further comprising:
triggering a predefined action that is associated with the specific thought within the thoughts data set.

7 . The method of claim 6 , wherein association of the predefined action with the specific thought within the thoughts data set is defined by a tuple that comprises a string data representation of the specific thought, the portion of the digital signal bitstream data identified as the thought chunk, a length value of the portion of the digital signal bitstream data identified as the thought chunk, a string data representation of the identified user category, string data descriptive of a demographic attribute of the person, and string data descriptive of the predefined action.

8 . The method of claim 7 , further comprising identifying the portion of the digital signal bitstream data as the thought chunk as a window portion of the analog electroencephalogram signals converted to the digital signal bitstream data digital signal bitstream data that is defined by a threshold changes in frequency domain values of the analog electroencephalogram signals over time at beginning and ending times of the window portion.

9 . The method of claim 7 , wherein the string data descriptive of a demographic attribute of the person is selected from the group consisting of a level of education of the person, a level of expertise of the person, and a school attended by the person.

10 . A system, comprising:
a processor; a computer readable memory in circuit communication with the processor; and a computer readable storage medium in circuit communication with the processor; wherein the processor executes program instructions stored on the computer-readable storage medium via the computer readable memory and thereby: converts analog electroencephalogram signals to a digital signal bitstream data, wherein the analog electroencephalogram signals are obtained from mental activity of a person; identifies a portion of the digital signal bitstream data as a thought chunk that represents discrete thought activity in response to matching, via a first artificial neural network comparison, first metadata of the thought chunk portion of the digital signal bitstream to metadata that is labeled in association with a thought within a thoughts data set, wherein the first artificial neural network is trained on the thoughts data set; identifies a user category to which the person generating the mental activity belongs in response to matching, via a second artificial neural network comparison, the first metadata and length and generation time metadata of the thought chunk portion to metadata that is labeled in association with a user category within the thoughts data set, wherein the second artificial neural network is trained on the thoughts data set and is different from the first artificial neural network; and identifies a specific thought within the thoughts data that has metadata that corresponds to the first metadata, the length and generation time metadata of the thought chunk portion, and to metadata that is labeled in association with the identified user category within the thoughts data set.

11 . The system of claim 10 , wherein the processor executes the program instructions stored on the computer-readable storage medium via the computer readable memory and thereby identifies the specific thought in response to matching, via a third artificial neural network comparison, the first metadata, the length and generation time metadata of the thought chunk portion and the metadata that is labeled in association with the identified user category to metadata that is labeled in association with the specific thought within the thoughts data set, wherein the third artificial neural network is trained on the thoughts data set and is different from the first artificial neural network and from the second artificial neural network.

12 .

Assignees

• IBM

Inventors

• Sgobba Nicolo'
• Rueger Erik
• Mattos Athila
• Takahashi Sergi
• Salnikov Eugen

Classifications

• A61B5/372

  Analysis of electroencephalograms · CPC title

• A61B5/7264 Primary

  Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems · CPC title

• G06F3/015 Primary

  Input arrangements based on nervous system activity detection, e.g. brain waves [EEG] detection, electromyograms [EMG] detection, electrodermal response detection · CPC title

• A61B5/7257

  using Fourier transforms · CPC title

• A61B5/7225

  Details of analogue processing, e.g. isolation amplifier, gain or sensitivity adjustment, filtering, baseline or drift compensation (input circuits for detecting, measuring, or recording bioelectric or biomagnetic signals A61B5/30; specific diagnostic methods using bioelectric or biomagnetic signals A61B5/316) · CPC title