Urvi Sharma

Title: Urvi Sharma and Machine-Learning Feature-Bias Evaluation

Introduction

Urvi Sharma is an inventor associated with Uttar Pradesh, India. Her listed patent work addresses methods for evaluating machine-learning models and detecting potential bias among feature classes.

Latest Patents

Sharma is associated with 1 patent, titled “Machine learning evaluation for detecting feature bias.” The invention describes an evaluation framework that receives evaluation data containing different classes of a feature and calculates at least two model-performance metrics. It then uses those metrics and their average to generate a unitless dissimilarity metric, reflecting variability associated with the feature. This approach is intended to help assess feature bias in machine-learning models.

Career Highlights

The provided information associates Sharma with Optum, Inc. Her listed patent highlights work on structured evaluation techniques for examining machine-learning performance across feature classes.

Collaborations

Amardeep Sharma is identified as a coworker. The available information does not specify her role in the patent or the nature of their collaboration.

Conclusion

Urvi Sharma’s listed invention focuses on measuring feature-related differences in machine-learning model performance, contributing an evaluation approach for detecting potential feature bias.