Radhika Agarwal

Title: Radhika Agarwal and Machine-Learning Patent Innovation.

Introduction

Radhika Agarwal is an inventor associated with West Bengal, India, and Optum, Inc. The supplied information lists 2 patents in her patent portfolio, including work on evaluating machine-learning models for feature bias.

Latest Patents

“Machine Learning Evaluation for Detecting Feature Bias” describes an evaluation framework for assessing whether machine-learning models perform differently across classes of an evaluation feature. The framework receives an evaluation dataset and uses an evaluation function to generate at least two performance metrics. It then calculates a unitless dissimilarity metric using those metrics and their average, providing a measure based on variability associated with the feature.

Career Highlights

Agarwal’s listed patent work focuses on methods for evaluating machine-learning systems. The disclosed approach offers a structured way to examine feature-related differences in model performance.

Collaborations

The supplied information identifies Optum, Inc. as Agarwal’s company and Amardeep Sharma as a coworker. It does not specify their roles in the patent or describe the nature of their collaboration.

Conclusion

Radhika Agarwal’s listed patent work addresses the evaluation of feature bias in machine-learning models, with an approach designed to quantify performance variability across feature classes.