Arun Abraham

Title: Arun Abraham and His Work in Deep Neural Network Quantization.

Introduction

Arun Abraham is an inventor associated with Karnataka, India. His patent work concerns methods for quantizing deep neural network models for use in electronic devices.

Latest Patents

Abraham is listed as an inventor on 1 patent, titled “Method and Electronic Device for Quantizing DNN Model.” The disclosed method estimates the activation range of each layer of a deep neural network using self-generated data, such as images, audio, or video, and/or a sensitivity index for each layer. It then quantizes the model based on the estimated activation ranges and/or sensitivity indices, assigning dynamic bit precision to each channel in each layer.

Career Highlights

Abraham is associated with Samsung Electronics. His patent reflects work on adapting quantization precision across neural network layers and channels, a technique intended to support the implementation of DNN models in electronic devices.

Collaborations

The listed coworkers associated with Abraham include Tejpratap Venkata Subbu Lakshmi Gollanapalli and Raja Kumar. The provided information does not specify their roles in the patent or the extent of their collaboration.

Conclusion

Arun Abraham’s recorded patent contribution focuses on dynamic, layer- and channel-level quantization of deep neural network models for electronic devices.