Mayank Rathee

Title: Innovations of Mayank Rathee in Deep Neural Networks

Introduction

Mayank Rathee is an accomplished inventor based in Panchkula, India. He has made significant contributions to the field of machine learning, particularly in the area of secure inference in deep neural networks. His innovative approach addresses critical privacy concerns in the deployment of artificial intelligence technologies.

Latest Patents

Mayank Rathee holds a patent titled "Private inference in deep neural network." This invention focuses on secure inference over Deep Neural Networks (DNNs) using secure two-party computation to perform privacy-preserving machine learning. The secure inference utilizes a specific type of comparison that can serve as a foundational element for various layers in the DNN, including ReLU activations and divisions. The method securely computes a Boolean share of a bit that indicates whether an input value x is less than another input value y. In this scenario, x is held by a user of the DNN, while y is held by a provider of the DNN. Each party's computing system parses their input into leaf strings of multiple bits, which is significantly more efficient than processing individual bits. This secure inference method is particularly well-suited for complex DNNs.

Career Highlights

Mayank Rathee is currently associated with Microsoft Technology Licensing, LLC, where he continues to push the boundaries of innovation in machine learning. His work is instrumental in developing technologies that enhance the security and privacy of data in neural networks.

Collaborations

Some of his notable coworkers include Nishanth Chandran and Divya Gupta, who contribute to the collaborative environment that fosters innovation at Microsoft.

Conclusion

Mayank Rathee's contributions to the field of deep neural networks exemplify the intersection of technology and privacy. His patent on secure inference represents a significant advancement in privacy-preserving machine learning.