Majid Jahani

Title: Majid Jahani: Innovator in Deep Learning Model Training Systems

Introduction

Majid Jahani is a notable inventor based in Bethlehem, PA, specializing in advancements in deep learning technologies. His innovative contributions have significantly impacted the field of artificial intelligence, particularly in the training of neural network models.

Latest Patents

Majid Jahani holds a patent for a "Deep learning model training system." This invention involves a comprehensive process where a computer trains a neural network model through several steps. Initially, observation vectors are randomly selected from a plurality of observation vectors. A forward and backward propagation of the neural network is executed to compute a gradient vector and a weight vector. Subsequently, a search direction vector and a step size value are computed. An updated weight vector is then calculated. Based on a predefined progress check frequency value, second observation vectors are randomly selected, and a progress check objective function value is computed using the weight vector, step size value, search direction vector, and second observation vectors. The mini-batch size value is updated based on an accuracy test. This iterative process continues until a convergence parameter indicates that the training of the neural network is complete.

Career Highlights

Majid Jahani is currently employed at SAS Institute Inc., where he continues to develop innovative solutions in the realm of data analytics and artificial intelligence. His work at SAS Institute Inc. allows him to collaborate with other talented professionals in the field.

Collaborations

Some of his notable coworkers include Joshua David Griffin and Seyedalireza Yektamaram, who contribute to the collaborative environment that fosters innovation and creativity.

Conclusion

Majid Jahani's contributions to deep learning model training systems exemplify the innovative spirit of modern technology. His work not only advances the field of artificial intelligence but also sets a foundation for future developments in neural network training methodologies.