Distributable model with biases contained within distributed data

Abstract

A system for improving a distributable model with biases contained in distributed data is provided, comprising a network-connected distributable model configured to serve instances of a plurality of distributable models; and a directed computation graph module configured to receive at least an instance of at least one of the distributable models from the network-connected computing system, create a cleansed dataset from data stored in the memory based at least in part by biases contained within the data stored in memory, train the instance of the distributable model with the cleansed dataset, and generate an update report based at least in part by updates to the instance of the distributable model.

What is claimed is: 
     
         1 . A system for improving a distributable model with biases contained in distributed data, comprising:
 a directed computation graph module comprising a memory, a processor, and a plurality of programming instructions stored in the memory thereof and operable on the processor thereof, wherein the programmable instructions, when operating on the processor, cause the processor to:
 receive a first distributable model from a distributable model source; 
 remove a plurality of data biases from the first distributable model using a plurality of data transformation nodes, wherein the biases comprise trends exhibited within the data; 
 create a generalized dataset based on the instance of the distributable model and the removed data biases for use in a second distributable model; 
 train the second distributable model with the generalized dataset; and 
 upload the second distributable model to the distributable model source; and 
 generate an update report based at least in part on changes made to the first distributable model to generate the second distributable model. 
   
     
     
         2 . The system of  claim 1 , wherein at least a portion of the generalized dataset is data that has had sensitive information removed. 
     
     
         3 . The system of  claim 1 , wherein the at least a portion of the update report is used by the distributable model source to improve the first distributable model. 
     
     
         4 . The system of  claim 1 , wherein at least a portion of the biases contained within the first distributable model is based on geography. 
     
     
         5 . The system of  claim 1 , wherein at least a portion of the biases contained within the first distributable model is based on age. 
     
     
         6 . The system of  claim 1 , wherein at least a portion of the biases contained within the first distributable model is based on gender. 
     
     
         7 . A method for improving a distributable model with biases contained in distributed data, comprising the steps of:
 (a) receiving a first distributable model from a distributable model source;   (b) removing a plurality of data biases within the first distributable model using a plurality of data transformation nodes in a directed computational graph module, wherein the biases comprise trends exhibited within the data;   (c) creating a generalized dataset based on the first distributable model and the removed biases for use in a second distributable model;   (d) training the second distributable model with the generalized dataset using the directed computation graph module;   (e) uploading the second distributable model to the distributable model source; and   (f) generating an update report based at least in part by updates to the instance of the distributable model with the directed computation graph module.   
     
     
         8 . The method of  claim 7 , wherein at least a portion of the generalized dataset is data that has had sensitive information removed. 
     
     
         9 . The method of  claim 7 , wherein the at least a portion of the update report is used by the distributable model source to improve the first distributable model. 
     
     
         10 . The method of  claim 7 , wherein at least a portion of the biases contained within the first distributable model is based on geography. 
     
     
         11 . The method of  claim 7 , wherein at least a portion of the biases contained within the first distributable model is based on age. 
     
     
         12 . The method of  claim 7 , wherein at least a portion of the biases contained within the first distributable model is based on gender.