17543149. AUTOMATED FAIRNESS-DRIVEN GRAPH NODE LABEL CLASSIFICATION simplified abstract (INTERNATIONAL BUSINESS MACHINES CORPORATION)
AUTOMATED FAIRNESS-DRIVEN GRAPH NODE LABEL CLASSIFICATION
Organization Name
INTERNATIONAL BUSINESS MACHINES CORPORATION
Inventor(s)
Ramasuri Narayanam of Bangalore (IN)
Sameep Mehta of Bangalore (IN)
Rakesh Rameshrao Pimplikar of Bangalore (IN)
Pranay Kumar Lohia of Bangalore (IN)
AUTOMATED FAIRNESS-DRIVEN GRAPH NODE LABEL CLASSIFICATION - A simplified explanation of the abstract
This abstract first appeared for US patent application 17543149 titled 'AUTOMATED FAIRNESS-DRIVEN GRAPH NODE LABEL CLASSIFICATION
Simplified Explanation
The patent application describes methods, systems, and computer program products for automated fairness-driven graph node label classification. This involves predicting node labels in a graph using a prediction model and generating an updated version of the model based on the predicted labels and fairness constraints. The updated model is then used to perform automated actions.
- Obtaining an input graph
- Predicting node labels using a graph node label prediction model
- Generating an updated version of the model based on predicted labels and fairness constraints
- Performing automated actions using the updated model
Potential Applications
- Social network analysis
- Recommender systems
- Fraud detection
- Image recognition
Problems Solved
- Ensuring fairness in graph node label classification
- Addressing biases and discrimination in automated decision-making systems
Benefits
- Improved fairness and equity in automated classification systems
- Reducing bias and discrimination in decision-making processes
- Enhancing the accuracy and reliability of graph node label classification
Original Abstract Submitted
Methods, systems, and computer program products for automated fairness-driven graph node label classification are provided herein. A computer-implemented method includes obtaining at least one input graph; predicting one or more node labels associated with the at least one input graph by processing at least a portion of the at least one input graph using a graph node label prediction model, wherein the graph node label prediction model includes at least one loss function; generating an updated version of the graph node label prediction model based at least in part on the one or more predicted node labels and one or more group fairness-based constraints relevant to the at least one input graph; and performing one or more automated actions using the updated version of the graph node label prediction model.