Thorsten Lorenzen

About

Thorsten Lorenzen is a leading figure in the field of efficient deep learning hardware, with a primary focus on VLSI architecture for convolutional neural network (CNN) inference. His most cited work, "CNN inference: VLSI architecture for convolution layer for 1.2 TOPS" (2017), addresses the critical challenge of deploying high-performance CNNs in resource-constrained environments such as automotive, industrial, and medical applications. Lorenzen’s major contribution lies in designing a specialized hardware architecture that achieves 1.2 trillion operations per second (TOPS) for the convolution layer—the computational bottleneck of modern CNNs—while maintaining energy efficiency and scalability. This work has garnered 6 citations, reflecting its foundational role in enabling real-time, low-power image classification on embedded systems. By bridging the gap between algorithmic advances in deep learning and practical hardware implementation, Lorenzen has helped pave the way for CNN deployment in autonomous vehicles, robotics, and medical diagnostics. His research continues to influence the development of application-specific integrated circuits (ASICs) for AI, making him a key contributor to the hardware acceleration ecosystem that powers today’s intelligent edge devices.

Research Focus

Key Achievements

Top Papers

1. 1
   CNN inference: VLSI architecture for convolution layer for 1.2 TOPS

   6 citations · 2017

Key Collaborators

Contact & Links