Abstract
The multidisciplinary nature of intelligent systems creates new challenges every day for researchers but also for companies all over the world. People face such systems in a lot of domains and real-life environments. Not only the development of theoretical aspects, but also practical applications have increasingly spread in the last decade to every field, from health to security or from agriculture to business management. Artificial intelligence, the development of novel, highly innovative theoretical methods and approaches, and also sophisticated applications and services are a focus of both academia and industry. Among numerous applications of artificial intelligence, image processing attracts great attention with the possibility of significant applications in a variety of domains. The combination of innovative algorithms to process massive image data are among the most important focal points of intelligent systems. In addition, innovative approaches in intelligent systems require discovering new problem-solving strategies, models, methodologies, and concurrent algorithms. This special issue of Concurrency and Computation: Practice and Experience includes extended high-quality papers presented at the IEEE International Conference on Innovations in Intelligent Systems and Applications (INISTA), which was held on August 24–26, 2020 in Novi Sad, Serbia, http://inista.org/inista20/index.php. Since 2004, the series of INISTA conferences has focused on intelligent systems, including both software and hardware, and provides a forum for researchers and industry to discuss new ideas and to exchange experiences with particular focus on innovative aspects and applications. INISTA 2020 attracted many international participants with over 80 submissions from all continents. Among them, 51 papers were accepted and published in the IEEE proceedings. Five papers have been selected and invited for this special issue. Authors were asked for extended papers containing at least 40% new material. After a further review process of extended versions of the conference papers, three were accepted to be published in this special issue. These papers focus on analyzing or developing new intelligent methods for object recognition, image segmentation, and building combinatorial algorithms for exploring large state-space graphs. Guney et al.1 have conducted a study named “Deep Neural Network Based Toddler Tracking System (Deep-TTS),” in which a system has been developed to warn parents if their children get close to any object that might be dangerous for them. In order to monitor the children, who have just learnt to walk, the face recognition FaceNet system has been used together with an object recognition algorithm named YOLOv3. The study performs the calculation of the Euclidean distance between the toddler and the dangerous object, where YOLOv3 is used to detect the object; and the parents are warned if this distance is below the threshold value. The system's test performance is observed to be over 90% as a result of the experimental studies. In the second article, Petrovic et al.2 present an empirical methodology based on nonadversarial perturbed datasets to analyze the sensitivity of deep learning methods for ocular fundus segmentation. The authors are interested in the effect of perturbations to the input images that might happen during normal image acquisition instead of adversarial attacks. Specifically, their focus is the problem of blur: Gaussian blur, approximating an unfocused image, and motion blur, approximating either the subject or the camera moving during capture. According to their analyzes, the architectures show larger variations in sensitivity to blur, and overfitting nonessential input dataset features and resolution sensitivity are a part of the problem. The third article3 “Exploring the Blocks World State Space,” by Bădică et al. is in the area of combinatorial algorithms for exploring large state-space graphs. In this article, the authors consider the blocks world, a prototype artificial intelligence problem often used to introduce problem solving strategies using searching, planning, and reasoning. The article presents algorithms for exploring, quantifying, and visualizing the state-space graph of the blocks world. The results include: edeclarative model of the blocks world state space graph, algorithms for evaluating several metrics on this graph (number of states, average number of stacks per state, number of transitions, and average branching factor), and experimental results involving proposed algorithms. Finally, we would like to thank all authors for their contributions to this special issue by extending their papers and the reviewers for their excellent job in reviewing the articles. We also extend our thanks to Professors David W. Walker, Jinjun Chen, Nitin Auluck and Martin Berzins, editors of the Concurrency and Computation: Practice and Experience, for offering us the opportunity to prepare this special issue. We hope that this special issue will attract the attention of readers interested in innovative issues of intelligent systems. Data sharing is not applicable to this article as no new data were created or analyzed in this study.