Boxborough, MA, United States of America

Marc M Hoffman

This inventor holds 24 USPTO granted patents and 2 published patent applications and 7 EPO patents, primarily in Neural Network Architecture. Top assignees: Qualcomm Incorporated, Analog Devices,inc., Intel Corporation. Active years: 2003-2026.

IDiyas Innovation Intelligence. (2026). Inventor Profile: Marc M Hoffman. Retrieved from https://idiyas.com/inventor/marc-m-hoffman

Data source: USPTO (United States Patent and Trademark Office) public patent records. Weekly synchronization. How IDiyas builds this profile

Calculating rank
USPTO Granted Patents = 24 

% Patents Active = 91.7

 

Average Co-Inventor Count = 3.8

ph-index = 6

Forward Citations = 107(Granted Patents)


Location History:

  • Boxborough, MA (US) (2014)
  • Mansfield, MA (US) (2003 - 2024)

Company Filing History:


Years Active: 2003-2026

Loading Chart...
Loading Chart...
Areas of Expertise:
Neural Networks
Quantization
Digital Signal Processing
Fast Fourier Transform
Data Extraction
Linear Activation Functions
Motion Estimation
Memory Bank
Vector Processor
Channel Estimation
Piecewise Linear Approximation
Addressing Mode
24 patents (USPTO):Explore Patents

Title: Innovations by Marc M. Hoffman: Pioneering Neural Network Technologies

Introduction

Marc M. Hoffman, an accomplished inventor based in Boxborough, MA, has made significant contributions to the field of technology with a remarkable portfolio of 21 patents. His groundbreaking work primarily focuses on advancing neural network operations and optimizing electronic device architectures.

Latest Patents

Among his latest innovations, Hoffman has developed two notable patents. The first patent, titled "Instruction Set Architecture for Neural Network Quantization and Packing," describes an electronic device capable of executing a single instruction to perform neural network operations on M-bit elements stored in input vector registers. This invention quantizes each element from M bits to P bits, where P is less than M, effectively generating a more efficient set of data for processing in neural networks.

The second patent, known as "Permutation Instruction," involves a device that includes a vector register file, a memory system, and a processor. This innovation enables the processor to access and execute a permutation instruction based on a periodicity parameter, which ensures efficient data source selection for populating vector registers.

Career Highlights

Marc M. Hoffman's professional journey includes significant roles at distinguished companies such as Analog Devices, Inc. and Qualcomm Incorporated. These experiences have undoubtedly contributed to his expertise and the innovative nature of his patents.

Collaborations

Throughout his career, Hoffman has collaborated with talented individuals, including Jose Fridman and Ajay Anant Ingle. These partnerships have fostered a collaborative environment that enhances creativity and drives technological advancements.

Conclusion

Marc M. Hoffman continues to be a leading figure in the realm of electronic devices and neural network technologies. His 21 patents reflect his dedication to innovation and his ability to rethink and enhance the functionality of technological systems. As advancements in neural networks and electronic architectures evolve, Hoffman's contributions are likely to have a lasting impact on the industry.

Profile summary based on public USPTO records.
Data Sources: USPTO Patent Grant XML, Patent Center, EPO & CIPO • Normalized by IDiyas Innovation Graph. Methodology & provenance architecturePlease report any incorrect information to [email protected]
Loading…