Bavaria, Germany

Martin Stefan Patz


Average Co-Inventor Count = 6.7

ph-index = 1


Location History:

  • Munich, DE (2020)
  • Bavaria, DE (2024)

Company Filing History:


Years Active: 2020-2024

Loading Chart...
3 patents (USPTO):Explore Patents

Title: Martin Stefan Patz: Innovator in Low-Power CNN Hardware Architecture

Introduction

Martin Stefan Patz is a distinguished inventor based in Bavaria, Germany, known for his innovative contributions in the realm of hardware architecture for convolutional neural networks (CNNs). With a remarkable portfolio of three patents, Patz continues to push the boundaries of technology, particularly in minimizing power consumption in advanced computing systems.

Latest Patents

One of his notable inventions is a patented low power hardware architecture for a convolutional neural network. This invention focuses on dynamic data quantization to effectively reduce the power requirements of systems implementing CNNs. The quantization scheme involves a compact representation of a 3×3 array of m-bit activation values using 9 n-bit mantissa values and a shared exponent. Additionally, the same quantization approach applies to a 3×3 kernel with p-bit parameter values. The functionality allows for efficient convolution of the kernel with activation data by computing the dot product between the mantissa values and summing the shared exponents. In CNNs that deploy multiple kernels, this architecture enables multiple computing units to operate simultaneously, optimizing performance while minimizing energy consumption.

Career Highlights

Patz currently serves as an inventor at Recogni Inc., where he dedicates his expertise to developing cutting-edge technology solutions. His work exemplifies a commitment to innovation, particularly in the fast-evolving field of artificial intelligence and machine learning.

Collaborations

Throughout his career, Martin has collaborated with notable professionals, including Shabarivas Abhiram and Eugene M Feinberg. These partnerships highlight the importance of teamwork in fostering innovation and advancing research within the tech community.

Conclusion

Martin Stefan Patz stands out as a key figure in the development of low-power hardware solutions for neural networks. His inventive spirit and dedication to reducing energy consumption in computing are instrumental in shaping the future of technology. As industries increasingly seek efficient and powerful solutions, innovators like Patz will continue to play a vital role.

This text is generated by artificial intelligence and may not be accurate.
Please report any incorrect information to support@idiyas.com
Loading…