We will present the possibilities of application and development, especially in the field of embedded systems, its interaction with other IoT components, the security aspects of individual components, as well as the domain of their interaction. In embedded systems, we will present new technologies such as fuzzy logic, application possibilities in embedded systems, and machine learning, as application possibilities through the implementation of machine learning. Then we will describe some more examples of the application of fuzzy logic, the automatic control of certain functions in cameras, as well as the defuzzification process, and the possibility of application in security cameras. In embedded systems, we will present the basic aspects of optimization and security, both in everyday applications and in interaction with other components of IoT technologies. The paper shows how security, reliability, and cost estimates affect the implementation phases and the final use of embedded systems, through examples of their application in industry. Security and data protection is shown through the construction of the mentioned devices, their application, but also different encryption methods, permissions, security devices at the network level, as well as implemented IoT technologies. Application examples are focused on the real segment, both in the field of transport, multimedia, design, and in the field of industrial application possibilities.

In research aimed at determining ways to protect the data of primary and secondary school students, as well as students and innovators who have submitted their ideas and innovations to innovation fairs in the territory of Republika Srpska, there is a lack of thoroughly analyzed methods and systems for protecting their ideas/innovations. This paper analyzes the most effective security algorithms for the protection of innovations and innovators from different categories. The objective of this work is to define the best prototype for protecting the identity database of innovators and innovations from the civil sector until their patent protection is granted in the territory of Bosnia and Herzegovina. By using the deductive method, we analyze various algorithms that function in a distributed environment. By comparing the advantages and disadvantages of existing algorithms, we suggest the application of the most appropriate one to meet the strategic decision-making needs of civil organizations.

This paper introduces a heuristic for multiple sequence alignment aimed at improving real-time object recognition in short video streams with uncertainties. It builds upon the idea of the progressive alignment but is cognitively economical to the extent that the underlying edit distance approach is adapted to account for human working memory limitations. Thus, the proposed heuristic procedure has a reduced computational complexity compared to optimal multiple sequence alignment. On the other hand, its relevance was experimentally confirmed. An extrinsic evaluation conducted in real-life settings demonstrated a significant improvement in number recognition accuracy in short video streams under uncertainties caused by noise and incompleteness. The second line of evaluation demonstrated that the proposed heuristic outperforms humans in the post-processing of recognition hypotheses. This indicates that it may be combined with state-of-the-art machine learning approaches, which are typically not tailored to the task of object sequence recognition from a limited number of frames of incomplete data recorded in a dynamic scene situation.

Cancelable biometrics is a demanding area of research in which a cancelable template conforming to a biometric is produced without degrading the efficiency. There are numerous approaches described in the literature that can be used to generate these cancelable templates. These approaches do not, however, perform well in either the qualitative or quantitative perspective. To address this challenge, a unique cancelable template generation mechanism based on signcryption and bio hash function is proposed in this paper. Signcryption is a lightweight cryptographic approach that uses hyper elliptic curve cryptography for encryption and a bio hash function for generating signatures in this proposed method. The cancelable templates are generated from iris biometrics. The hybrid grey level distancing method is used for perfect iris feature extraction for the CASIA and IITD datasets. The proposed approach is compared against the existing state-of-the-art cancelable techniques. The resulting analysis reveals that the proposed method is efficient in terms of accuracy of 98.86%, with lower EER of 0.1%. The average minimum TPR and TNR of the proposed method is about 99.81%.

The purpose of this article is to introduce to the literature a new extension of the Simple WISP method adapted for utilizing the triangular fuzzy numbers. This extension is proposed to allow the use of the Simple WISP method for addressing decision-making problems related to uncertainties and inaccuracies, as well as for solving problems related to predictions. In addition, this article also discusses the use of linguistic variables to collect the attitudes of the respondents, as well as their transformation into appropriate triangular fuzzy numbers. The article discusses the use of two defuzzification procedures. The first normalization procedure is easy to use, while the second procedure uses the advantages that the application of asymmetric fuzzy numbers gives in terms of analysis. The usability of the proposed extension is presented through two examples.

This paper describes some of the forms of violence in cyberspace and possible prevention. The second part of this paper provides an analysis of students’ views on possible online abuse and prevention. The research aims to highlight statistically significant opportunities for raising awareness of possible online abuse and the prevention of cyberbullying. The survey consists of a total of 20 testimonials, which are divided into 4 parts, and each part assesses different aspects. The conclusion of this research is that the Internet environment does not provide enough feedback on the reactions of the person to whom the harassing messages were sent, which in the abuser reduces the feeling of causing true emotional and psychological harm to another person, reducing the degree of self-control and insight into the level of bullying.

In the paper, the possibility of combining deep neural network (DNN) model compression methods to achieve better compression results was considered. To compare the advantages and disadvantages of each method, all methods were applied to the ResNet18 model for pretraining to the NCT-CRC-HE-100K dataset while using CRC-VAL-HE-7K as the validation dataset. In the proposed method, quantization, pruning, weight clustering, QAT (quantization-aware training), preserve cluster QAT (hereinafter PCQAT), and distillation were performed for the compression of ResNet18. The final evaluation of the obtained models was carried out on a Raspberry Pi 4 device using the validation dataset. The greatest model compression result on the disk was achieved by applying the PCQAT method, whose application led to a reduction in size of the initial model by as much as 45 times, whereas the greatest model acceleration result was achieved via distillation on the MobileNetV2 model. All methods led to the compression of the initial size of the model, with a slight loss in the model accuracy or an increase in the model accuracy in the case of QAT and weight clustering. INT8 quantization and knowledge distillation also led to a significant decrease in the model execution time.

This paper is dedicated to machine learning, the branches of machine learning, which include the methods for solving this issue, and the practical implementation of the solution to the automatic image description generation. Automatic image caption generation is one of the frequent goals of computer vision. Image description generation models must solve a larger number of complex problems to have this task successfully solved. The objects in the image must be detected and recognized, after which a logical and syntactically correct textual description is generated. For that reason, description generation is a complex problem. It is an extremely important challenge for machine learning algorithms because it represents an impersonation of a complicated human ability to encapsulate huge amounts of highlighted visual pieces of information in descriptive language. The results of the generated descriptions are compared depending on the used pretrained convolutional networks. The BLEU metrics are used to calculate the quality of the image description. Although the solution to the problem of image description automatic generation does provide us with good results, there is yet room for improvement since there are images that are not adequately described.

Ayurvedic medicines are categorized into seven constitutional forms ‘Prakriti’ which is a constituent in the Ayurvedic system of medicine to determine drought tolerance and drug responsiveness. Prakriti assessment entails a thorough physical examination as well as queries about physiological or behavioral characteristics. The prevalence of certain "doshas" is attributed by Ayurveda to the fundamental constituent of a person. Vata, pitta, and Kapha are the three main doshas mentioned. Ayurveda-dosha studies have been used for a long time, but the quantitative reliability measurement of these diagnostic methods still lags. The careful and appropriate analysis leads to an effective treatment. In this paper, we demonstrate the result of certain machine learning methods like Support Vector Machine (SVM), Naive Bayes (NB), Decision Tree (DT), K-Nearest Neighbour (KNN), Artificial Neural Network (ANN), and Adaboost algorithm for various performance characteristics to predict human body constituencies. From the observations of results it is shown that the AdaBoost algorithm with hyperparameter tuning provides enhanced accuracy and recall of 0.97, precision and F-score of 0.96, the lower RSME value obtained is 0.64. The experimental results reveal that the improved model, which is based on ensemble learning methods, outperforms traditional methods significantly. According to the findings, advancements in the proposed algorithms could give machine learning a promising future.