Presentation information

Interactive Session

[3Rin4] Interactive 1

Thu. Jun 11, 2020 1:40 PM - 3:20 PM Room R01 (jsai2020online-2-33)

[3Rin4-19] Adaptive Distributed Compressed Sensing with Cooperative Multi-Agent Deep Reinforcement Learning

〇Masatoshi Sekine1, Satoshi Ikada1 (1.Oki Electric Industry Co., Ltd.)

Keywords:multi-agent, deep reinforcement learning, compressed sensing, distributed cooperative control, sensor networks

In recent years, monitoring systems using IoT for factory equipment and infrastructure have become important. To improve the transmission efficiency of sensor data collected over a wide area for a long period of time, we have developed LACSLE (Lightweight and Adaptive Compressed Sensing based on deep Learning for Edge devices). LACSLE is a machine learning-based adaptive compression sensing method that dynamically estimates the optimal compression ratio according to the data. In this paper, we propose a more efficient compressed sensing method based on multi-agent deep reinforcement learning to extend LACSLE to the cooperation of multiple terminals. By a basic study of our proposed method, it was found that more efficient compression and reconstruction can be performed by adaptive distributed compressed sensing. In addition, we found that the actions of compression and reconstruction patterns can be expressed by one-dimensional continuous values respectively in the framework of cooperative multi-agent reinforcement learning.

Authentication for paper PDF access

A password is required to view paper PDFs. If you are a registered participant, please log on the site from Participant Log In.
You could view the PDF with entering the PDF viewing password bellow.