Journal:Informatica
Volume 34, Issue 3 (2023), pp. 577–602
Abstract
Healthcare has seen many advances in sensor technology, but with recent improvements in networks and the addition of the Internet of Things, it is even more promising. Current solutions to managing healthcare data with cloud computing may be unreliable at the most critical moments. High response latency, large volumes of data, and security are the main issues of this approach. The promising solution is fog computing, which offers an immediate response resistant to disconnections and ways to process big data using real-time analytics and artificial intelligence (AI). However, fog computing has not yet matured and there are still many challenges. This article presents for a computer scientist a systematic review of the literature on fog computing in healthcare. Articles published in six years are analysed from the service, software, hardware, information technologies and mobility with autonomy perspectives. The contribution of this study includes an analysis of recent trends, focus areas and benefits of the use of AI techniques in fog computing e-health applications.
Journal:Informatica
Volume 34, Issue 4 (2023), pp. 771–794
Abstract
The quality of the input data is amongst the decisive factors affecting the speed and effectiveness of recurrent neural network (RNN) learning. We present here a novel methodology to select optimal training data (those with the highest learning capacity) by approaching the problem from a decision making point of view. The key idea, which underpins the design of the mathematical structure that supports the selection, is to define first a binary relation that gives preference to inputs with higher estimator abilities. The Von Newman Morgenstern theorem (VNM), a cornerstone of decision theory, is then applied to determine the level of efficiency of the training dataset based on the probability of success derived from a purpose-designed framework based on Markov networks. To the best of the author’s knowledge, this is the first time that this result has been applied to data selection tasks. Hence, it is shown that Markov Networks, mainly known as generative models, can successfully participate in discriminative tasks when used in conjunction with the VNM theorem.
The simplicity of our design allows the selection to be carried out alongside the training. Hence, since learning progresses with only the optimal inputs, the data noise gradually disappears: the result is an improvement in the performance while minimising the likelihood of overfitting.
Journal:Informatica
Volume 34, Issue 3 (2023), pp. 491–527
Abstract
Embedding models turn words/documents into real-number vectors via co-occurrence data from unrelated texts. Crafting domain-specific embeddings from general corpora with limited domain vocabulary is challenging. Existing solutions retrain models on small domain datasets, overlooking potential of gathering rich in-domain texts. We exploit Named Entity Recognition and Doc2Vec for autonomous in-domain corpus creation. Our experiments compare models from general and in-domain corpora, highlighting that domain-specific training attains the best outcome.
Journal:Informatica
Volume 34, Issue 3 (2023), pp. 465–489
Abstract
The Best-Worst Method (BWM) is a recently introduced, innovative multi-criteria decision-making (MCDM) technique used to determine criterion weights for selection processes. However, another method is needed to complete the selection of the most preferred alternative. In this research, we propose a group decision-making methodology based on the multiplicative BWM to make this selection. Furthermore, we give new models that allow for groups with different best and worst criteria to exist. This capability is crucial in reconciling the differences among experts from various geographical locations with diverse evaluation perspectives influenced by social and cultural disparities. Our work contributes significantly in three ways: (1) we propose a BWM-based methodology for evaluating alternatives, (2) we present new linear models that facilitate decision-making for groups with different best and worst criteria, and (3) we develop a dissimilarity ratio to quantify the differences in expert opinions. The methodology is illustrated via numerical experiments for a global car company deciding which car model alternative to introduce in its markets.
Journal:Informatica
Volume 34, Issue 3 (2023), pp. 665–677
Abstract
Due to the complexity and lack of transparency of recent advances in artificial intelligence, Explainable AI (XAI) emerged as a solution to enable the development of causal image-based models. This study examines shadow detection across several fields, including computer vision and visual effects. Three-fold approaches were used to construct a diverse dataset, integrate structural causal models with shadow detection, and apply interventions simultaneously for detection and inferences. While confounding factors have only a minimal impact on cause identification, this study illustrates how shadow detection enhances understanding of both causal inference and confounding variables.
Pub. online:28 Aug 2023Type:Research ArticleOpen Access
Journal:Informatica
Volume 34, Issue 3 (2023), pp. 529–556
Abstract
Ineffective evaluation of open-source software learning management system (OSS-LMS) packages can negatively impact organizational effectiveness. Clients may struggle to select the best OSS-LMS package from a wide range of options, leading to a complex multi-criteria group decision-making (MCGDM) problem. This evaluates OSS-LMS packages based on several criteria like usability, functionality, e-learning standards, reliability, activity tracking, course development, assessment, backup and recovery, error reporting, efficiency, operating system compatibility, computer-managed instruction, authentication, authorization, troubleshooting, maintenance, upgrading, and scalability. Handling uncertain data is a vital aspect of OSS-LMS package evaluation. To tackle MCGDM issues, this study presents a consensus weighted sum product (c-WASPAS) method which is applied to an educational OSS-LMS package selection problem to evaluate four OSS-LMS packages, namely ATutor, eFront, Moodle, and Sakai. The findings indicate that the priority order of alternatives is Moodle > Sakai > eFront > ATutor and, therefore, MOODLE is the best OSS-LMS package for the case study. A sensitivity analysis of criteria weights is also conducted, as well as a comparative study, to demonstrate the effectiveness of the proposed method. It is essential to note that proper OSS-LMS package evaluation is crucial to avoid negative impacts on organizational performance. By addressing MCGDM issues and dealing with uncertain information, the c-WASPAS method presented in this study can assist clients in selecting the most appropriate OSS-LMS package from multiple alternatives. The findings of this study can benefit educational institutions and other organizations that rely on OSS-LMS packages to run their operations.
Pub. online:25 Aug 2023Type:Research ArticleOpen Access
Journal:Informatica
Volume 34, Issue 3 (2023), pp. 635–663
Abstract
This paper focuses on games on augmenting systems with a coalition structure that can be seen as an extension of games with a coalition structure and games on augmenting systems. Considering the player payoffs, the quasi-Owen value is defined. To show the rationality of this payoff index, five representative axiomatic systems are established. The population monotonic allocation scheme (PMAS) and the core are introduced. Moreover, the relationships between the PMAS and quasi-Owen value as well as the core and quasi-Owen value are discussed. Finally, an illustrative example is given to show the concrete application of the new payoff indices.
Pub. online:28 Jun 2023Type:Research ArticleOpen Access
Journal:Informatica
Volume 34, Issue 3 (2023), pp. 557–576
Abstract
The widespread use of sensors has resulted in an unprecedented amount of time series data. Time series mining has experienced a particular surge of interest, among which, subsequence matching is one of the most primary problem that serves as a foundation for many time series data mining techniques, such as anomaly detection and classification. In literature there exist many works to study this problem. However, in many real applications, it is uneasy for users to accurately and clearly elaborate the query intuition with a single query sequence. Consequently, in this paper, we address this issue by allowing users to submit a small query set, instead of a single query. The multiple queries can embody the query intuition better. In particular, we first propose a novel probability-based representation of the query set. A common segmentation is generated which can approximate the queries well, in which each segment is described by some features. For each feature, the corresponding values of multiple queries are represented as a Gaussian distribution. Then, based on the representation, we design a novel distance function to measure the similarity of one subsequence to the multiple queries. Also, we propose a breadth-first search strategy to find out similar subsequences. We have conducted extensive experiments on both synthetic and real datasets, and the results verify the superiority of our approach.
Pub. online:15 Jun 2023Type:Research ArticleOpen Access
Journal:Informatica
Volume 34, Issue 3 (2023), pp. 449–464
Abstract
Many confidential multimedia, such as personal privacy, commercial, and military secrets, are transmitted on the Internet. To prevent this confidential multimedia from being eavesdropped on by illegal users, information-hiding technology is a leading research topic nowadays. One of the important research topics of information-hiding technology is coverless information hiding, which utilizes the unchanged property of its multimedia carrier to hide secret information. In this paper, we propose two schemes that employ the average pixel value of an image. The first is an extension of the Coverless Information Hiding Based on the Most Significant Bit (CIHMSB) scheme, referred to as E-CIHMSB. In the E-CIHMSB, we build an extended matrix containing the image fragment’s average pixel value. The second scheme is a combination theory-based CIHMSB, referred to as CB-CIHMSB. In the CB-CIHMSB, we construct the combined matrix. E-CIHMSB and CB-CIHMSB embed the secret bits by changing the most significant bits of the chosen pixel in the matrix. Experimental results show that our schemes achieved higher hiding capacity than previous related schemes. Moreover, the proposed scheme is more robust against steganalysis tools and image quality attacks such as Additive Gaussian White Noise (AWGN), Salt & Pepper noise, low-pass filtering attacks, and JPEG compression attacks than CIHMSB.
Pub. online:15 Jun 2023Type:Research ArticleOpen Access
Journal:Informatica
Volume 34, Issue 2 (2023), pp. 285–315
Abstract
Over the past decades, many methods have been proposed to solve the linear or nonlinear mixing of spectra inside the hyperspectral data. Due to a relatively low spatial resolution of hyperspectral imaging, each image pixel may contain spectra from multiple materials. In turn, hyperspectral unmixing is finding these materials and their abundances. A few main approaches to performing hyperspectral unmixing have emerged, such as nonnegative matrix factorization (NMF), linear mixture modelling (LMM), and, most recently, autoencoder networks. These methods use different approaches in finding the endmember and abundance of information from hyperspectral images. However, due to the huge variation of hyperspectral data being used, it is difficult to determine which methods perform sufficiently on which datasets and if they can generalize on any input data to solve hyperspectral unmixing problems. By trying to mitigate this problem, we propose a hyperspectral unmixing algorithm testing methodology and create a standard benchmark to test already available and newly created algorithms. A few different experiments were created, and a variety of hyperspectral datasets in this benchmark were used to compare openly available algorithms and to determine the best-performing ones.