Sanghamitra Dutta

Title: **Inventor Spotlight: Sanghamitra Dutta**

Introduction

Sanghamitra Dutta, located in Yorktown Heights, NY, is an innovative inventor known for his contributions to machine learning. He has made significant strides in the field through his patented technology, which enhances the efficiency of distributed stochastic gradient descent.

Latest Patents

Dutta holds a patent titled "Adaptive Learning Rate Schedule in Distributed Stochastic Gradient Descent." This method focuses on optimizing machine learning processes by assigning processing jobs to multiple model learners while utilizing a central parameter server. The technique involves solving gradients based on a current set of parameters, iterating these parameters as results are returned. A critical aspect of this patent is determining the "staleness" of gradient solving, which informs how learning rates are adjusted, ultimately improving the adaptability and performance of machine learning systems.

Career Highlights

Sanghamitra Dutta is associated with the International Business Machines Corporation (IBM), a leading company in technology and innovation. His work at IBM involves collaborating with cutting-edge technologies and developing methods that advance the field of artificial intelligence and machine learning.

Collaborations

During his tenure at IBM, Dutta has collaborated with notable coworkers such as Parijat Dube and Gauri Joshi. These collaborations have enabled him to leverage diverse expertise and contribute to groundbreaking research and development in the technology sector.

Conclusion

Sanghamitra Dutta's contributions to machine learning through his patented invention reflect his commitment to innovation. His work continues to influence the field, paving the way for more advanced and efficient machine learning techniques. As a part of IBM, he remains at the forefront of technological advancements, fostering collaborations that drive the future of AI.