Seyed Sajad Mousavi

Title: Seyed Sajad Mousavi: Innovator in Machine Learning Robustness

Introduction

Seyed Sajad Mousavi is a prominent inventor based in Milpitas, California. He has made significant contributions to the field of machine learning, particularly in enhancing the robustness of image classification models. His innovative approach utilizes reinforcement learning to address challenges in black-box image classification.

Latest Patents

Mousavi holds a patent for a "Reinforcement learning agent to measure robustness of black-box image classification models." This patent outlines systems and methods for reinforcement learning agents that conduct adversarial black-box attacks. The goal is to determine and refine the robustness of machine learning models. The process involves receiving an image corresponding to a ground truth and computing the sensitivity of the model in classifying the image against added and removed distortions. The reinforcement learning agent intelligently decides which distortions to add or remove based on these sensitivities. The machine learning model then classifies the image, and this iterative process continues until a misclassification occurs. This misclassification serves as a basis for determining robustness and can lead to retraining the machine learning model.

Career Highlights

Mousavi is currently employed at Hewlett Packard Enterprise Development LP, where he continues to push the boundaries of machine learning technology. His work focuses on developing innovative solutions that enhance the performance and reliability of machine learning systems.

Collaborations

Some of his notable coworkers include Soumyendu Sarkar and Ashwin Ramesh Babu. Their collaborative efforts contribute to the advancement of technology in the field of machine learning.

Conclusion

Seyed Sajad Mousavi is a key figure in the innovation of machine learning robustness. His patent and work at Hewlett Packard Enterprise Development LP highlight his commitment to advancing technology in this critical area.