Extending machine learning training data to generate an artificial intelligence recommendation engine

Abstract

A catalog of physical items associated with a target user is accessed. At least a portion of the catalog is at least in part automatically generated based on a retention of one or more of the physical items provided to the target user. A machine learning model trained using outfit combination information gathered from other users is used to automatically determine for the target user, at least a portion of one or more recommended outfit combinations of a plurality of physical items among the physical items within the catalog. An indication of a selected one of the one or more recommended outfit combinations is provided to the target user.

What is claimed is: 
     
         1 . A method, comprising:
 accessing, by one or more processors, a catalog of physical items associated with a target user, wherein at least a portion of the catalog is at least in part automatically generated based on a retention of one or more of the physical items provided to the target user;   training, by the one or more processors, a machine learning model using a training set of data associated with a segmented target category, wherein a plurality of users are included in the segmented target category based on a plurality of user attributes, wherein training the machine learning model includes adding or modifying one or more user attributes to the plurality of user attributes until a threshold amount of training data is included in the training set of data associated with the segmented target category, wherein the training set of data associated with the segmented target category includes using outfit combination information gathered from other users;   using, by the one or more processors, the trained machine learning model to automatically determine for the target user, at least a portion of one or more recommended outfit combinations of a plurality of physical items among the physical items within the catalog;   providing to a reviewer a listing of the one or more recommended outfit combinations;   receiving from the reviewer an identification of a subset of the one or more recommended outfit combinations to be provided to the target user; and   providing the identified subset of the one or more recommended outfit combinations to the target user.   
     
     
         2 . The method of  claim 1 , wherein the outfit combination information gathered from the other users is a selected subset among a larger set of available outfit combination information for a group of users that includes at least the other users. 
     
     
         3 . The method of  claim 1 , wherein the outfit combination information gathered from the other users is selected for use in training the machine learning model including by identifying one or more defining features of the target user and determining the other users that share the one or more defining features. 
     
     
         4 . The method of  claim 1 , wherein the machine learning model is one of a plurality of available machine learning models and the machine learning model is selected for use based on a user segment corresponding to the target user. 
     
     
         5 . The method of  claim 4 , wherein each of the plurality of available machine learning models corresponds to different user segments. 
     
     
         6 . The method of  claim 1 , further comprising receiving a feedback of the identified subset of the one or more recommended outfit combinations from the target user. 
     
     
         7 . The method of  claim 6 , wherein the feedback includes an outfit combination style preference of the target user. 
     
     
         8 . The method of  claim 6 , wherein the feedback includes a description of a modified outfit combination based on the identified subset of the one or more recommended outfit combinations. 
     
     
         9 . The method of  claim 1 , further comprising:
 receiving from the target user a submission describing one or more additional physical items; and   updating the catalog of physical items associated with the target user to include the one or more additional physical items.   
     
     
         10 . The method of  claim 1 , further comprising receiving a weather context for the target user, wherein the one or more recommended outfit combinations are automatically determined based at least in part on the received weather context. 
     
     
         11 . The method of  claim 1 , further comprising receiving one or more shared calendar events of the target user, wherein the one or more recommended outfit combinations are automatically determined based at least in part on the received one or more shared calendar events. 
     
     
         12 . The method of  claim 11 , wherein the one or more shared calendar events include a wedding, a business meeting, a vacation, or an exercise class. 
     
     
         13 . The method of  claim 1 , further comprising receiving a specification of a recently worn item by the target user, wherein the recommended outfit combinations are automatically determined based at least in part on excluding the recently worn item from the catalog of physical items associated with the target user until a time threshold has elapsed. 
     
     
         14 . The method of  claim 1 , wherein a delivery time of the providing the identified subset of the one or more recommended outfit combinations is configured by the target user. 
     
     
         15 . The method of  claim 1 , further comprising generating a packing list of physical items corresponding to the identified subset of the one or more recommended outfit combinations. 
     
     
         16 . A system, comprising:
 one or more processors configured to:
 access a catalog of physical items associated with a target user, wherein at least a portion of the catalog is at least in part automatically generated based on a retention of one or more of the physical items provided to the target user; 
 train, a machine learning model using a training set of data associated with a segmented target category, wherein a plurality of users are included in the segmented target category based on a plurality of user attributes, wherein training the machine learning model includes adding or modifying one or more user attributes to the plurality of user attributes until a threshold amount of training data is included in the training set of data associated with the segmented target category, wherein the training set of data associated with the segmented target category includes using outfit combination information gathered from other users; 
 use the trained machine learning model to automatically determine for the target user, at least a portion of one or more recommended outfit combinations of a plurality of physical items among the physical items within the catalog; 
 provide to a reviewer a listing of the one or more recommended outfit combinations; 
 receive from the reviewer an identification of a subset of the one or more recommended outfit combinations to be provided to the target user; and 
 provide the identified subset of the one or more recommended outfit combinations to the target user; and 
   a memory coupled to the one or more processors and configured to provide the one or more processors with instructions.   
     
     
         17 . The system of  claim 16 , wherein the outfit combination information gathered from the other users is a selected subset among a larger set of available outfit combination information for a group of users that includes at least the other users. 
     
     
         18 . The system of  claim 16 , wherein the outfit combination information gathered from the other users is selected for use in training the machine learning model including by identifying one or more defining features of the target user and determining the other users that share the one or more defining features. 
     
     
         19 . The system of  claim 16 , wherein the machine learning model is one of a plurality of available machine learning models and the machine learning model is selected for use based on a user segment corresponding to the target user. 
     
     
         20 . A computer program product embodied in a non-transitory computer readable medium and comprising computer instructions for:
 accessing, by one or more processors, a catalog of physical items associated with a target user, wherein at least a portion of the catalog is at least in part automatically generated based on a retention of one or more of the physical items provided to the target user;   training, by the processor, a machine learning model using a training set of data associated with a segmented target category, wherein a plurality of users are included in the segmented target category based on a plurality of user attributes, wherein training the machine learning model includes adding or modifying one or more user attributes to the plurality of user attributes until a threshold amount of training data is included in the training set of data associated with the segmented target category, wherein the training set of data associated with the segmented target category includes using outfit combination information gathered from other users;   using, by the one or more processors, the trained machine learning model to automatically determine for the target user, at least a portion of one or more recommended outfit combinations of a plurality of physical items among the physical items within the catalog;   providing to a reviewer a listing of the one or more recommended outfit combinations;   receiving from the reviewer an identification of a subset of the one or more recommended outfit combinations to be provided to the target user; and   providing the identified subset of the one or more recommended outfit combinations to the target user.