Martin Power

Title: Martin Power: Innovator in Neural Network Acceleration

Introduction

Martin Power is a distinguished inventor based in Dublin, Ireland. He has made significant contributions to the field of neural network accelerators, holding a total of seven patents. His work focuses on enhancing the efficiency and performance of machine learning models through innovative storage and processing techniques.

Latest Patents

One of Martin Power's latest patents is titled "Methods and apparatus for sparse tensor storage for neural network accelerators." This invention discloses methods, apparatus, systems, and articles of manufacture for sparse tensor storage. An example apparatus includes circuitry for generating a sparsity map that indicates whether a data point of the tensor is zero. It also features static storage controlling circuitry that divides the tensor into storage elements and a compressor that performs two levels of compression to optimize storage.

Another notable patent is "Methods and apparatus to perform machine-learning model operations on sparse accelerators." This invention outlines methods and systems to execute machine-learning model operations efficiently. The apparatus includes circuitry that generates sparsity data based on acceleration operations and instructs data buffers to provide activation or weight data accordingly.

Career Highlights

Martin Power has worked with prominent companies in the technology sector, including Intel Corporation and Movidius Limited. His experience in these organizations has allowed him to develop and refine his innovative ideas in neural network technology.

Collaborations

Throughout his career, Martin has collaborated with talented individuals such as Martin-Thomas Grymel and Niall Hanrahan. These collaborations have contributed to the advancement of his projects and the successful development of his patents.

Conclusion

Martin Power is a notable inventor whose work in neural network acceleration has led to several impactful patents. His innovative approaches to sparse tensor storage and machine-learning operations continue to influence the field of artificial intelligence.