Bias reduction in machine learning model training and inference

Abstract

One or more default protected attribute values may be determined for a prediction model trained based on training data including a plurality of training observations. Each of the plurality of training observations may include a respective plurality of training data values corresponding with a plurality of features. Each of the plurality of training observations may also include a respective target value. Each of the plurality of training observations may include a respective protected attribute value corresponding with a protected attribute feature. A request to determine a designated predicted target value for a designated inference observation may be received after determining the one or more default protected attribute values. The predicted target value may be determined by applying the prediction model to an inference observation and a designated default protected attribute value of the one or more default protected attribute values.

1 . A method comprising:
 determining one or more default protected attribute values for a prediction model trained based on training data including a plurality of training observations, each of the plurality of training observations including a respective plurality of training data values corresponding with a plurality of features, each of the plurality of training observations also including a respective target value, each of the plurality of training observations including a respective protected attribute value corresponding with a protected attribute feature;   receiving via a communication interface a request to determine a designated predicted target value for a designated inference observation after determining the one or more default protected attribute values, the designated inference observation including a designated plurality of inference data values corresponding with the plurality of features;   determining the designated predicted target value via a processor by applying the prediction model to the designated inference observation and a designated default protected attribute value of the one or more default protected attribute values; and   storing the predicted target value on a storage device.   
     
     
         2 . The method recited in  claim 1 , the method further comprising:
 determining via the processor a plurality of predicted target values including the designated predicted target value by applying the prediction model to the designated default protected attribute value and a plurality of inference observations including the designated inference observation, each of the plurality of inference observations including a respective plurality of inference data values corresponding with the plurality of features.   
     
     
         3 . The method recited in  claim 1 , the method further comprising:
 determining a plurality of evaluation metric values indicating performance of the prediction model for each of a plurality of candidate default protected attribute values, wherein the one or more default protected attribute values are determined at least in part based on the plurality of evaluation metrics.   
     
     
         4 . The method recited in  claim 1 , wherein determining the one or more default protected attribute values involves determining an overlap profile between the protected attribute feature and a designated feature of the plurality of features, the overlap profile indicating a respective degree of overlap among the plurality of training observations between first selected values corresponding to the protected attribute feature and second selected values corresponding to the designated feature. 
     
     
         5 . The method recited in  claim 4 , the method further comprising:
 determining based on the overlap profile that a designated one of the respective degrees of overlap indicates a positivity violation; and   identifying one or more value replacement rules for correcting the positivity violation by replacing a feature value or a protected attribute value.   
     
     
         6 . The method recited in  claim 5 , the method further comprising:
 determining a replacement data value based on the one or more value replacement rules; and   replacing an original feature value or a protected attribute value in the inference observation with the replacement data value.   
     
     
         7 . The method recited in  claim 1 , wherein the prediction model is a regression model that includes a plurality of regression coefficients corresponding with the plurality of features, a designated one or more of the plurality of regression coefficients corresponding with the protected attribute feature, wherein applying the prediction model to the inference observation involves determining a constant term based on the designated default protected attribute value and the designated one or more regression coefficients. 
     
     
         8 . The method recited in  claim 1 , wherein the prediction model is a neural network that includes a plurality of neurons corresponding with the plurality of features, a designated one of the plurality of neurons corresponding with the protected attribute feature, wherein applying the prediction model to the inference observation involves determining a constant value for the designated neuron based on the designated default protected attribute value. 
     
     
         9 . The method recited in  claim 1 , wherein the prediction model is selected from the group consisting of: a tree-based model, a neural network model, and a gradient boosting model. 
     
     
         10 . The method recited in  claim 1 , wherein each of the training observations corresponds to a respective individual, and wherein the protected attribute is selected from the group consisting of: race, ethnicity, sex, gender, national origin, religion, disability status, age, genetic information, marital status, and receipt of public assistance. 
     
     
         11 . One or more non-transitory computer-readable media having instructions stored thereon for performing a method, the method comprising:
 determining one or more default protected attribute values for a prediction model trained based on training data including a plurality of training observations, each of the plurality of training observations including a respective plurality of training data values corresponding with a plurality of features, each of the plurality of training observations also including a respective target value, each of the plurality of training observations including a respective protected attribute value corresponding with a protected attribute feature;   receiving via a communication interface a request to determine a designated predicted target value for a designated inference observation after determining the one or more default protected attribute values, the designated inference observation including a designated plurality of inference data values corresponding with the plurality of features;   determining the designated predicted target value via a processor by applying the prediction model to the designated inference observation and a designated default protected attribute value of the one or more default protected attribute values; and   storing the predicted target value on a storage device.   
     
     
         12 . The one or more non-transitory computer-readable media recited in  claim 11 , the method further comprising:
 determining via the processor a plurality of predicted target values including the designated predicted target value by applying the prediction model to the designated default protected attribute value and a plurality of inference observations including the designated inference observation, each of the plurality of inference observations including a respective plurality of inference data values corresponding with the plurality of features.   
     
     
         13 . The one or more non-transitory computer-readable media recited in  claim 11 , the method further comprising:
 determining a plurality of evaluation metric values indicating performance of the prediction model for each of a plurality of candidate default protected attribute values, wherein the one or more default protected attribute values are determined at least in part based on the plurality of evaluation metrics.   
     
     
         14 . The one or more non-transitory computer-readable media recited in  claim 11 , wherein the prediction model is a regression model that includes a plurality of regression coefficients corresponding with the plurality of features, a designated one or more of the plurality of regression coefficients corresponding with the protected attribute feature, wherein applying the prediction model to the inference observation involves determining a constant term based on the designated default protected attribute value and the designated one or more regression coefficients. 
     
     
         15 . The one or more non-transitory computer-readable media recited in  claim 11 , wherein the prediction model is a neural network that includes a plurality of neurons corresponding with the plurality of features, a designated one of the plurality of neurons corresponding with the protected attribute feature, wherein applying the prediction model to the inference observation involves determining a constant value for the designated neuron based on the designated default protected attribute value. 
     
     
         16 . The one or more non-transitory computer-readable media recited in  claim 11 , wherein each of the training observations corresponds to a respective individual, and wherein the protected attribute is selected from the group consisting of: race, ethnicity, sex, gender, national origin, religion, disability status, age, genetic information, marital status, and receipt of public assistance. 
     
     
         17 . A method comprising:
 determining via a processor a prediction model based on training data including a plurality of training observations, each of the plurality of training observations including a respective plurality of training data values corresponding with a first plurality of features, each of the plurality of training observations also including a respective target value, wherein the first plurality of features includes a protected attribute;   determining via a processor a designated predicted target value by applying the prediction model to a designated inference observation including a plurality of inference data values corresponding with a second plurality of features, wherein the second plurality of features excludes the protected attribute; and   storing the predicted target value on a storage device.   
     
     
         18 . The method recited in  claim 17 , the method further comprising:
 determining via the processor a plurality of predicted target values including the designated predicted target value by applying the prediction model to a plurality of inference observations including the designated inference observation, each of the plurality of inference observations including a respective plurality of inference data values corresponding with the second plurality of features.   
     
     
         19 . The method recited in  claim 17 , wherein the prediction model is a regression model that includes a plurality of regression coefficients, some or all of the plurality of regression coefficients corresponding with the first plurality of features, wherein the plurality of regression coefficients includes a designated one or more coefficients corresponding with the protected attribute, and wherein the designated one or more coefficients are omitted from the regression model when determining the predicted target value. 
     
     
         20 . The method recited in  claim 17 , wherein each of the training observations corresponds to a respective individual, and wherein the protected attribute is selected from the group consisting of: race, ethnicity, sex, gender, sexual orientation, national origin, religion, disability status, age, genetic information, marital status, and receipt of public assistance.