Deepika Bablani

Title: Deepika Bablani: Innovator in Artificial Neural Networks

Introduction

Deepika Bablani is a prominent inventor based in San Jose, California. She has made significant contributions to the field of artificial intelligence, particularly in the area of artificial neural networks. Her innovative work has led to the development of a unique patent that enhances the functionality of neural networks.

Latest Patents

Deepika holds a patent titled "Learned Step Size Quantization." This invention provides a method for quantizing input values in artificial neural networks. The system comprises an artificial neural network with a quantizer that has a configurable step size. This quantizer receives multiple input values and quantizes them according to the configurable step size, producing a set of quantized input values. The network also includes a matrix multiplier that applies weights to these quantized values to generate output values with a specified precision. Additionally, a multiplier scales these output values to a different precision. This innovative approach optimizes the training process of neural networks by adjusting the step size based on gradients and training input data.

Career Highlights

Deepika is currently employed at International Business Machines Corporation (IBM), where she continues to push the boundaries of technology. Her work at IBM has allowed her to collaborate with some of the brightest minds in the industry, contributing to advancements in artificial intelligence and machine learning.

Collaborations

Some of her notable coworkers include Steve Kyle Esser and Jeffrey L McKinstry. Their collaborative efforts have further enhanced the research and development of innovative technologies within the company.

Conclusion

Deepika Bablani's contributions to artificial neural networks exemplify her dedication to innovation in technology. Her patent on learned step size quantization is a testament to her expertise and commitment to advancing the field of artificial intelligence.