Abhishek Kumar

Title: Abhishek Kumar and His Work in Multiple Kernel Learning.

Introduction

Abhishek Kumar is an inventor associated with College Park, Maryland, and Nec Laboratories America, Inc. His patent work addresses methods for making multiple kernel learning more effective and scalable.

Latest Patents

Kumar is credited with 1 patent: “Two-stage multiple kernel learning method.” The invention frames multiple kernel learning as a standard binary classification problem while adding constraints to ensure that the learned kernel remains positive definite. This approach enables the use of binary classification technologies to develop multiple kernel learning methods that are simpler, more scalable, and potentially better performing.

Career Highlights

Kumar’s listed invention reflects work at the intersection of machine learning and classification. The method’s focus on scalability and conceptual simplicity offers an alternative framework for developing multiple kernel learning systems.

Collaborations

Kumar’s listed coworkers include Koray Kavukcuoglu and Alexandru Niculescu-Mizil.

Conclusion

Abhishek Kumar’s patent presents a constrained binary-classification framework for multiple kernel learning, highlighting an approach designed to improve scalability and performance.