Custom Hardware Inference Accelerator for TensorFlow Lite for Microcontrollers

Article Computer Science, Information Systems

Using HW/SW Codesign for Deep Neural Network Hardware Accelerator Targeting Low-Resources Embedded Processors

Erez Manor et al.

Summary: The usage of RISC-based embedded processors, which provide low-cost and low-power solutions, is increasing in popularity. By utilizing hardware acceleration and Instruction Set Architecture (ISA) extension, high performance yet low power embedded processors can be achieved. Efficient mapping of computational load onto hardware and software resources is a key challenge for improving performance while maintaining low power and area. This paper proposes a graph analysis-based methodology for efficient hardware/software partitioning, which offers a solution for achieving overall speedup and area requirements.

IEEE ACCESS (2022)

Add to Collection

Article Chemistry, Analytical

Quantization and Deployment of Deep Neural Networks on Microcontrollers

Pierre-Emmanuel Novac et al.

Summary: Embedding Artificial Intelligence onto low-power devices is a challenging task that has seen some progress in recent years but still has room for optimization. This study introduces a new framework called MicroAI for quantizing and deploying deep neural networks onto low-power microcontrollers, and provides a comparison study with existing embedded inference engines.

SENSORS (2021)

Add to Collection

Review Chemistry, Analytical

An Overview of Machine Learning within Embedded and Mobile Devices-Optimizations and Applications

Taiwo Samuel Ajani et al.

Summary: Embedded systems technology is undergoing transformation due to advancements in computer architecture and machine learning applications. However, implementing machine learning algorithms in resource-constrained environments like embedded and mobile devices requires innovative optimization techniques. Current research trends focus on optimizing computationally and memory-intensive algorithms for embedded and mobile computing environments, with an emphasis on key application areas and future exploration in embedded machine learning.

SENSORS (2021)

Add to Collection

Article Computer Science, Information Systems