There is a vast consensus in the embedded system community about the need to support artificial intelligence at the edge. An increasing demand for IoT and automotive devices is experienced with such a mandatory need to provide additional value to the user. Pushing further the need at its extreme, the device becomes a sensing element integrated in a ultra tiny package. To stimulate further progress in that respect a challenge has been set to the 2023 edition of the IEEE COINS conference. This paper describes this challenge as well as its results. It was aimed to let international teams use an end-to-end novel methodology to automatically conceive and deploy machine learning solutions for human activity recognition, as exemplary case study. The target tiny device was an intelligent sensor processor equipped with programmable capabilities to deploy neuton.ai automatically generated machine learning workloads. The contest winning team devised a tiny neural model which achieved balanced accuracy of 81.4% over 24 classes on neuton.ai deep learning framework. The model was deployed on the sensor, clocked at 10 MHz, and required a total memory of 19 KiB and its inference latency was 152 ms.

In the context of energy transmission, Line Rating addresses the capacity of conductors to safely transmit energy under different conditions. Two distinct rating methods are known: Static Line Rating and Dynamic Line Rating, cor?responding to fixed worst-case weather conditions and real?time variable conditions, respectively. The assumption underlying Static Line Rating, that conductor temperature and line sag are deterministically correlated, is challenged by real-world findings, revealing stochastic relationships influenced by uncontrollable weather variability. The potential for safely operating trans?mission lines beyond their static-rated capacity is discussed, emphasizing the use of Dynamic Line Rating with advanced technologies. A novel architecture involving temperature sensors, laser diodes, and optical fibers is introduced for achieving Dynamic Line Rating. The proposed system enables real-time temperature monitoring and transmission to a control station for optimized energy transport. Various implementation options, from integrated solutions to external attachments, are considered. The abstract concludes with an overview of the learning process for modeling temperature behaviors and line current allocation using an Echo State Network, showcasing its suitability for capturing dynamic complexities in the context of transmission line rating.

This paper sets its primary objective to  understand the transformative potential of Tiny Machine  Learning (Tiny ML) and Artificial Intelligence (AI) in  enhancing industrial efficiency. Using a research design that  juxtaposes the theoretical understanding of these technologies  with real-world applications, the methodology adopted  emphasizes three pivotal use cases, substantiated with tangible  examples. The main outcomes show the influential role of  STMicroelectronics, a leading semiconductor entity, in bridging  the gap between Tiny ML, AI, and industrial applications.  Results from in-depth examinations highlight the value of  Predictive Maintenance as evidenced by offshore wind farms,  the importance of Gesture Recognition in Human-Machine  Interaction with a focus on autonomous vehicles, and the  efficient integration of Tiny ML into IoT Sensor Networks,  notably seismic monitoring. In conclusion, this manuscript underscores the impending future where Tiny ML and AI  synergize, particularly hinting at breakthroughs in humanoid  robotics. Such advancements are anticipated to redefine the  contours of human-technology interaction impacting everyone  in everyday life.

TinyRCE is a hyperspherical classifier aimed at Continual Learning On-Tiny-Devices, a challenging task in which a Machine Learning model is required to learn from continuous streams of data while being directly installed on a (tiny) device with limited computational resources. The classifier has so far been applied to several use cases, including Human Activity Recognition, Ball Bearing Anomaly Classification, Keyword Spotting and Image Classification. The proposed work in this paper focuses on the reproducibility of TinyRCE’s experimental results already published on other papers. This to prove that all the published results are quantitatively reproducible. All the experiments have been executed on two independent computing machines to profile the impact on accuracy of the computations. As the outcomes are matching, the experimental reproducibility of TinyRCE’s accuracy over all the use cases has been positively verified.

This paper presents the state-of-the-art review of the different approaches for Neural Architecture Search targeting resource constrained devices such as microcontrollers. As well as the implementations of On-Device learning techniques for those devices.