Location History:
- Santa Clara, CA (US) (2020)
- Durham, NC (US) (2022 - 2023)
Company Filing History:
Years Active: 2020-2023
Title: Innovations by Georgios Mappouras
Introduction
Georgios Mappouras is an accomplished inventor based in Santa Clara, CA. He has made significant contributions to the field of neural networks and memory management, holding a total of 3 patents. His work focuses on enhancing the efficiency and reliability of memory systems in computing.
Latest Patents
Mappouras's latest patents include innovative solutions for neural network training and memory management. One of his notable inventions is a system for runtime extension for neural network training with heterogeneous memory. This system includes a neural network connected to both a low-capacity, high-bandwidth memory and a high-capacity, low-bandwidth memory. During the forward propagation pass, a run-time manager monitors buffer usage across various layers. In the backward propagation pass, it optimizes buffer movement between the two types of memory, effectively reducing memory access latency.
Another significant patent addresses tolerating memory stack failures in multi-stack systems. This invention improves the reliability of memory stacks by allowing the system to recover from failures without data loss. A front-end controller divides data into sub-blocks and creates a redundant reliability sub-block. This enables the system to detect and recover from errors in memory stacks, ensuring continued operation even in the event of a failure.
Career Highlights
Georgios Mappouras is currently employed at Advanced Micro Devices Corporation, where he continues to innovate in the field of memory management and neural networks. His work has been instrumental in advancing the capabilities of modern computing systems.
Collaborations
Mappouras collaborates with talented individuals such as Amin Farmahini Farahani and Nuwan S Jayasena, contributing to a dynamic and innovative work environment.
Conclusion
Georgios Mappouras is a notable inventor whose work in neural networks and memory management has led to significant advancements in technology. His patents reflect a commitment to improving system efficiency and reliability, making him a valuable asset in the field of computing.