Habituation-compensated library of affective response

• US 8,918,344 B2
• Filed: 06/25/2011
• Issued: 12/23/2014
• Est. Priority Date: 05/11/2011
• Status: Active Grant

• Current Assignee: Affectomatics ltd.
• Original Assignee: Anat Thieberger Ben-Haim, Ari M. Frank, Gil Thieberger
• Inventors: Frank, Ari M., Thieberger, Gil, Thieberger Ben-Haim, Anat
• Primary Examiner(s): Gaffin, Jeffrey A
• Assistant Examiner(s): FINK, THOMAS ANDREW
• Application Number:
  US13/168,964
• Publication Number:
  US 20120290513A1
• Time in Patent Office: 1,277 Days
• Field of Search: None
• US Class Current: 706/12
• CPC Class Codes: G06F 3/013 Eye tracking input arrangements G06F3/015 takes precedence
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/02 Knowledge representation; Symbolic representation

Abstract

Generating a habituation-compensated library comprising a user'"'"'s expected response to tokens representing stimuli that influence the user'"'"'s affective state, the method comprising: receiving samples comprising temporal windows of token instances to which the user was exposed, wherein the token instances have overlapping instantiation periods; the samples further comprise data on previous instantiations of at least one of the token instances from the temporal windows; receiving target values corresponding to the temporal windows of token instances; the target values represent the user'"'"'s response to the token instances from the temporal windows of token instances; training a machine learning-based user response model using the samples, the data on previous instantiations, and the corresponding target values; and analyzing the machine learning-based user response model to generate the habituation-compensated library, which accounts for the influence of the user'"'"'s previous exposure to tokens.

Claims

1. A method for generating a habituation-compensated library, comprising:

receiving samples comprising temporal windows of token instances to which a user was exposed, wherein the temporal windows of token instances comprise a window comprising instantiations of first and second tokens that have overlapping instantiation periods;

receiving data on previous instantiations of the first and second tokens, to which the user was exposed;

receiving target values corresponding to the temporal windows of token instances;

the target values represent affective responses of the user to the token instances from the temporal windows of token instances;

wherein the affective responses are values comprising representations of emotional responses;

training a machine learning-based user response model using data comprising;

the samples, the data on previous instantiations of the first and second tokens, and the corresponding target values; and

generating, based on the machine learning-based user response model, the habituation-compensated library that comprises for each token of the first and second tokens;

a first expected affective response of the user to an instance of the token after a first number of previous exposures to instantiations of the token, and a second expected affective response of the user to an instance of the token after a second number, that is greater than the first number, of previous exposures to instantiations of the token;

wherein for the first token, the first expected affective response is stronger than the second expected affective response, while for the second token, the first expected affective response is weaker than the second expected affective response.

2. The method of claim 1, wherein the habituation-compensated library comprises expected emotional responses of the user to instantiations of tokens.

3. The method of claim 1, wherein the data on previous instantiations comprises at least one of:

elapsed since last time the user was exposed to an instantiation of the at least one token, and a number of previous instantiations of the at least one token the user was exposed to.

4. The method of claim 1, wherein the habituation-compensated library is generated by analyzing at least two different machine learning-based user response models that were trained on data collected over periods during which the user was in different situations;

and the habituation-compensated library comprises expected affective responses of the user to same tokens in different situations.

5. The method of claim 1, wherein the step of training the machine learning-based user response model comprises training a neural network;

and the habituation-compensated library comprises partial derivatives of output values of the neural network, with respect to input neurons that correspond to token instances.

6. The method of claim 1, wherein the step of training the machine learning-based user response model comprises training a regression model;

and the habituation-compensated library comprises regression parameters.

7. The method of claim 1, wherein the step of training the machine learning-based user response model comprises training a maximum entropy model classifier;

and the habituation-compensated library comprises feature function weighting parameters.

17. The method of claim 2, further comprising predicting affective response annotations from user measurement channel values of the user;

wherein the affective response annotations are expressed as emotional responses; and

further comprising utilizing the affective response annotations as target values.

8. A method for generating a habituation-compensated library, comprising:

receiving samples comprising temporal windows of token instances to which a user was exposed, wherein the temporal windows of token instances comprise a window comprising instantiations of first and second tokens that have overlapping instantiation periods;

receiving data on previous instantiations of the first and second tokens, to which the user was exposed;

receiving target values corresponding to the temporal windows of token instances;

the target values, which are derived from values of a measurement channel of the user, represent affective responses of the user to the token instances from the temporal windows of token instances;

wherein the affective responses are values comprising representations of values of the measurement channel of the user;

training a machine learning-based user response model using data comprising;

the samples, the data on previous instantiations of the first and second tokens, and the corresponding target values; and

generating, based on the machine learning-based user response model, the habituation-compensated library that comprises for each token of the first and second tokens;

a first expected affective response of the user to an instance of the token after a first number of previous exposures to instantiations of the token, and a second expected affective response of the user to an instance of the token after a second number, that is greater than the first number, of previous exposures to instantiations of the token;

wherein for the first token, the first expected affective response is stronger than the second expected affective response, while for the second token, the first expected affective response is weaker than the second expected affective response.

9. The method of claim 8, wherein the data on previous instantiations comprises at least one of:

time elapsed since last time the user was exposed to an instantiation of the at least one token, and a number of previous instantiations of the at least one token the user was exposed to.

10. The method of claim 8, wherein the habituation-compensated library is generated by analyzing at least two different machine learning-based user response models that were trained on data collected over periods during which the user was in different situations;

and the habituation-compensated library comprises expected affective responses of the user to same tokens in different situations.

11. The method of claim 8, wherein the step of training the machine learning-based user response model comprises training a neural network;

and the habituation-compensated library comprises partial derivatives of output values of the neural network, with respect to input neurons that correspond to token instances.

12. The method of claim 8, wherein the step of training the machine learning-based user response model comprises training a regression model;

and the habituation-compensated library comprises regression parameters.

13. The method of claim 8, wherein the step of training the machine learning-based user response model comprises training a maximum entropy model classifier;

and the habituation-compensated library comprises feature function weighting parameters.

18. The method of claim 8, wherein the habituation-compensated library comprises expected affective responses of the user to instantiations of tokens;

wherein the affective responses are expressed as values of a user measurement channel of the user.

14. A device comprising a processor and a memory;

the memory configured to store samples and target values;

the samples comprising temporal windows of token instances to which a user was exposed, wherein the temporal windows of token instances comprise a window comprising instantiations of first and second tokens that have overlapping instantiation periods;

the samples further comprise data on previous instantiations of the first and second tokens, to which the user was exposed; and

the target values correspond to the temporal windows of token instances and represent affective responses of the user to the token instances from the temporal windows of token instances;

wherein the affective responses comprise at least one of;

values representing emotional responses, and values of a user measurement channel of the user;

the processor configured to train a machine learning-based user response model using date comprising;

the samples, the data on previous instantiations of the first and second tokens, and the corresponding target values stored in the memory; and

the processor is further configured to generate, based on the machine learning-based user response model, a habituation-compensated library that comprises for each token of the first and second tokens;

a first expected affective response of the user to an instance of the token after a first number of previous exposures to instantiations of the token, and a second expected affective response of the user to an instance of the token after a second number, that is greater than the first number, of previous exposures to instantiations of the token;

wherein for the first token, the first expected affective response is stronger than the second expected affective response, while for the second token, the first expected affective response is weaker than the second expected affective response.

15. The device of claim 14, wherein the habituation-compensated library comprises expected emotional responses of the user to instantiations of tokens.

16. The device of claim 15, further comprising a target value generator configured to receive user measurement channel values of the user and to predict target values that are affective response annotations from the user measurement channels values of the user;

wherein the affective response annotations are expressed as emotional responses.