Abstract
Missing data is a problem often found in real-world datasets and it can degrade the performance of most machine learning models. Several deep learning techniques have been used to address this issue, and one of them is the Autoencoder and its Denoising and Variational variants. These models are able to learn a representation of the data with missing values and generate plausible new ones to replace them. This study surveys the use of Autoencoders for the imputation of tabular data and considers 26 works published between 2014 and 2020. The analysis is mainly focused on discussing patterns and recommendations for the architecture, hyperparameters and training settings of the network, while providing a detailed discussion of the results obtained by Autoencoders when compared to other state-of-the-art methods, and of the data contexts where they have been applied. The conclusions include a set of recommendations for the technical settings of the network, and show that Denoising Autoencoders outperform their competitors, particularly the often used statistical methods.

Abstract
The Internet of Things (IoT) is rapidly becoming an integral component of the industrial market in areas such as automation and analytics, giving rise to what is termed as the Industrial IoT (IIoT). The IIoT promises innovative business models in various industrial domains by providing ubiquitous connectivity, efficient data analytics tools, and better decision support systems for a better market competitiveness. However, IIoT deployments are vulnerable to a variety of security threats at various levels of the connectivity and communications infrastructure. The complex nature of the IIoT infrastructure means that availability, confidentiality and integrity are difficult to guarantee, leading to a potential distrust in the network operations and concerns of loss of critical infrastructure, compromised safety of network end-users and privacy breaches on sensitive information. This work attempts to look at the requirements currently specified for a secure IIoT ecosystem in industry standards, such as Industrial Internet Consortium (IIC) and OpenFog Consortium, and to what extent current IIoT connectivity protocols and platforms hold up to the standards with regard to security and privacy. The paper also discusses possible future research directions to enhance the security, privacy and safety of the IIoT.

Abstract
AbstractIn this position paper, our goal is to argue the need for specific software development practices to early-stage startups. In order to reach this goal, we discuss the consequences of innovative and market-driven contexts, which are two of the key elements when describing software startups. We also argue that these practices could be applied to innovative initiatives within established companies since they share similar characteristics and challenges as those from startups.

Abstract
Nowadays, industrial robots are still commonly programmed using essentially off-line tools, such as is the case of structured languages or simulated environments. This is a very time-consuming process, which necessarily requires the presence of an experienced programmer with technical knowledge of the set-up to be used, as well as a concept and a complete definition of the details associated with the operations. Moreover, considering some industrial applications such as coating, painting, and polishing, which commonly require the presence of highly skilled shop floor operators, the translation of this human craftsmanship into robot language using the available programming tools is still a very difficult task. In this regard, this paper presents a programming by demonstration solution, that allows a skilled shop floor operator to directly teach the industrial robot. The proposed system is based on the 6D Mimic innovative solution, endowed with an IMU sensor as to enable the system to tolerate temporary occlusions of the 6D Marker. Results show that, in the event of an occlusion, a reliable and highly accurate pose estimation is achieved using the IMU data. Furthermore, the selected IMU was a low-cost model, to not severely increase the 6D Mimic cost, despite lowering the quality of the readings. Even in these conditions, the developed algorithm was able to produce high-quality estimations during short time occlusions.

Abstract
Medical training in oral health includes interpreting images of the oral cavity. The use of extended datasets, real cases, and the ability to monitor the progress of students is beneficial. In this work, we investigate the feasibility of using a teledentistry solution to support medical training. The rOral integrated teledentistry solution, developed by the research team, allows the decoupling of image acquisition and image review: images of the oral cavity can be recorded using the camera of regular mobile devices, near the subject of care; the images are then uploaded to a secure storage and reviewed by experts, using a web environment. This workflow, primarily designed for remote oral health diagnosis (e.g.: population screening), can be used for dentistry education, allowing students to play the role of experts. A population of dentistry students was involved in the present study and used rOral to access anonymized cases to form a diagnosis. The results show that the existing solution, planned for remote diagnosis, is suitable for education and that smartphone-acquired images can be used in diagnosis assessment activities. © 2020, Springer Nature Switzerland AG.

Abstract
Background: The occurrence of an orofacial trauma can originate health, social, economic and professional problems. A 13-year boy suffered the avulsion of tooth 11 and 21, lost at the scenario. Methods: Three intraoral appliances were manufactured: A Hawley appliance with a central expansion screw and two central incisors (1), trumpet edentulous anterior tooth appliance (2) and a customized splint (3) were designed as part of the rehabilitation procedure. Objectively assessing the sound quality of the trumpet player with these new devices in terms of its spectral, temporal, and spectro-temporal audio properties. A linear frequency response microphone was adopted for precision measurement of pitch, loudness, and timbre descriptors. Results: Pitch deviations may result from the different intra-oral appliances due to the alteration of the mouth cavity, respectively, the area occupied and modification/interaction with the anatomy. This investigation supports the findings that the intra-oral appliance which occupies less volume is the best solution in terms of sound quality. Conclusions: Young wind instrumentalists should have dental impressions of their teeth made, so their dentist has the most reliable anatomy of the natural teeth in case of an orofacial trauma. Likewise, the registration of their sound quality should be done regularly to have standard parameters for comparison.