Ayaz Mehmani

Title: Ayaz Mehmani and His Work on Artificial Intelligence Training Data

Introduction

Ayaz Mehmani is an inventor based in McLean, Virginia. His patent work addresses a challenge in artificial intelligence: how to train models when the underlying training data shifts over time.

Latest Patents

Mehmani is associated with 1 patent, titled “Systems and Methods for Mitigating Training Data Baseline Shifts in Artificial Intelligence Models Comprising Moving Classifications.” The invention organizes training data into multiple classes according to baseline shifts and trains a model or ensemble using the combined dataset, rather than requiring separate training for each class. It also generates user-specific input features from profile information, such as auto loan terms or credit history, and adds features based on time criteria. These criteria help determine how the model processes the input, including the data’s originating class and its overlap with relevant time periods.

Career Highlights

Mehmani is associated with Capital One Services, LLC. His patent reflects an application of machine-learning methods to data that changes over time, with potential relevance to financial profiles and related modeling tasks.

Collaborations

Ruoyu Shao is identified as Mehmani’s coworker. The available information does not specify whether Shao collaborated on the patent.

Conclusion

Ayaz Mehmani’s patent focuses on adapting artificial intelligence models to shifting training-data baselines through classification, time-based features, and integrated model training.