Systems and methods for detecting and regularizing drift in sequential decisions

A method to identify drift in a dynamic context of real-world decision-making, the method comprising:

• receiving, from a user-computing device, a search query from a user;
• determining, by a processor, whether a plurality of history drift data for the user is available;
• when the plurality of history drift data for the user is available, providing, by the processor, a plurality of items in a spatial layout that the user may choose for a predetermined duration and at a predetermined frequency;
• displaying, on the user-computing device, the plurality of items in the spatial layout for the predetermined duration and at the predetermined frequency;
• monitoring, by the processor via at least one sensor, an eye gaze of the user while making sequential decisions about the plurality of items;
• determining, by the processor, that a drift has occurred;
• regularizing, by the processor, a plurality of alternative choice options when drift is detected;
• learning and identifying, by the processor, an upper boundary and a lower boundary based on whether the user chooses or rejects at least one alternative choice of the plurality of alternative choice options;
• modeling, by the processor, the drift using machine learning based at least in part on a history of user decision responses;
• predicting, by the processor, a drift direction of the drift and a relevance impact of different attributes into a drift state;
• classifying, by the processor, the drift states into a risk-averse classification and a risk-seeking classification by clustering drift parameters over different behavioral measures;
• and
• outputting, by the processor, a desirable behavior model for machine learning.

A system to identify drift in a dynamic context of real-world decision-making, the system comprising:

• a user-computing device;
• at least one sensor communicatively coupled to the user-computing device, the at least one sensor captures data as an input signal for a temporal drift;
• a processor;
• and
• a non-transitory, processor-readable storage medium in communication with the processor, the non-transitory, processor-readable storage medium comprising one or more programming instructions that, when executed, cause the processor to:
• receive a search query from a user;
• determine whether a plurality of history drift data for the user is available;
• when the plurality of history drift data for the user is available, provide a plurality of items in a spatial layout that the user may choose for a predetermined duration and at a predetermined frequency;
• output for display on the user-computing device a plurality of items in a spatial layout that the user may choose for a predetermined duration and at a predetermined frequency;
• monitor an eye gaze of the user while making sequential decisions about the plurality of items;
• determine that a drift has occurred;
• regularize a plurality of alternative choice options when drift is detected;
• learn and identify an upper boundary and a lower boundary based on whether the user chooses or rejects at least one alternative choice of the plurality of alternative choice options;
• model the drift using machine learning based at least in part on a history of user decision responses;
• predict a drift direction of the drift and a relevance impact of different attributes into a drift state;
• classify the drift states into a risk-averse classification and a risk-seeking classification by clustering drift parameters over different behavioral measures to learn an attentive drift of the user;
• and
• output a desirable behavior model for machine learning,
• wherein the prediction of the drift states, the classifying of the drift states, and the modeling of the drift provide the data for training the desirable behavior model for machine learning.