Aniruddha Saha

Title: Aniruddha Saha: Innovator in Adversarial Machine Learning

Introduction

Aniruddha Saha is a notable inventor based in Elkridge, MD (US). He has made significant contributions to the field of machine learning, particularly in developing systems that enhance the robustness of models against adversarial attacks. With a total of three patents to his name, Saha is recognized for his innovative approaches to image processing and defense mechanisms.

Latest Patents

Saha's latest patents include a "System and method with masking for certified defenses against adversarial patches." This invention relates to a computer-implemented system that generates a set of one-mask images from a source image. It employs a first mask across predetermined image regions to ensure comprehensive coverage of the source image. The system selects a one-mask image with the highest prediction loss and generates a set of two-mask images for further training of the machine learning system.

Another significant patent is the "System and method with masking and inpainting strategy for generic defense against patch attacks." This invention involves dividing a source image into multiple regions and using a mask to create a first masked region. The method generates a protected image based on similarity assessments between the masked and inpainted regions, enhancing the image's resilience against potential attacks.

Career Highlights

Aniruddha Saha is currently employed at Robert Bosch GmbH, where he applies his expertise in machine learning and image processing. His work focuses on developing innovative solutions that address the challenges posed by adversarial machine learning.

Collaborations

Saha collaborates with talented individuals such as Chaithanya Kumar Mummadi and Shuhua Yu, contributing to a dynamic research environment that fosters innovation and creativity.

Conclusion

Aniruddha Saha's contributions to the field of machine learning through his patents demonstrate his commitment to advancing technology and improving system defenses against adversarial threats. His work continues to influence the landscape of image processing and machine learning.