A Review of Artificial Intelligence in Embedded Systems

Round 1

Reviewer 1 Report

The authors provide a comprehensive review of the existing literature with adequate tables and figures.

The quality of the English language is high. 

Reviewer 2 Report

The paper is written to sufficiently present three aspects of methods and applications for deploying artificial intelligence (AI) models and networks on embedded devices. Such is a review of artificial intelligence in Embedded Systems. The article offers three aspects of methods and applications for deploying AI technologies on embedded devices being solvable.  

The paper fits the body of knowledge of AI and Embedded Systems with views of reducing the complexity of many machine learning models. The report proposed a solution by reviewing the literature based on classification and associated references.  The proposal anchor on the zero-activation-skipping convolutional accelerator to avoid the non-contributing multiplication of zero-valued activations.  Although, the sub-modules could be further expanded to simplify computational intelligence techniques for image analysis adopted by the authors. 

The paper provided linkages to further research in the current deployment state in embedded AI from the three perspectives in the review.  

The authors could further clarify the need for three perspectives and the further development of efficient algorithms, methods, and applications for the optimization of hardware acceleration methods.  However, the authors underscore the outlook of Embedded AI and summarise the expected clarification.  For these reasons, the overall merit of the article is high.