Search

Trust plays a pivotal role in numerous humans’ interactions, and it affects relationships ranging from individuals to interstate levels. Any conflict and its resolution can therefore benefit from insights into trust dynamics. The deployment of computational trust management methods and agent technologies provide promising options. They have already been successfully applied in many domains, while in conflict research (and possible resolution policies) they are rather unexplored. This research paper therefore presents an approach that is based on a formalized, anthropocentric trust management method. It deploys it accordingly in agents’ simulations to obtain new insights into the dynamics of trust in cooperation–conflict interplay, and to enable derivation of strategies for managing societies to drive them toward desired states.

Knowledge about the interactions between dietary and biomedical factors is scattered throughout uncountable research articles in an unstructured form (e.g., text, images, etc.) and requires automatic structuring so that it can be provided to medical professionals in a suitable format. Various biomedical knowledge graphs exist, however, they require further extension with relations between food and biomedical entities. In this study, we evaluate the performance of three state-of-the-art relation-mining pipelines (FooDis, FoodChem and ChemDis) which extract relations between food, chemical and disease entities from textual data. We perform two case studies, where relations were automatically extracted by the pipelines and validated by domain experts. The results show that the pipelines can extract relations with an average precision around 70%, making new discoveries available to domain experts with reduced human effort, since the domain experts should only evaluate the results, instead of finding, and reading all new scientific papers.

Paving the way towards the realization of Mark Weiser's vision of ubiquitous computing [1], the research community has made incredible advancements on several fronts. When it comes to interacting with humans, for example, computers can already use pretty much anything as a touchpad [2]. Similarly, when it comes to sensing the environment, computers can unobtrusively detect anything from a driver fatigue [3] to the presence of the queen bee in a hive [4]. When compared with these, advancements on the "core" front - the computing itself - appear to be rather orthodox and limited.

Purpose Existing research on the measurability of information quality (IQ) has delivered poor results and demonstrated low inter-rater agreement measured by Intra-Class Correlation (ICC) in evaluating IQ dimensions. Low ICC could result in a questionable interpretation of IQ. The purpose of this paper is to analyse whether assessors’ motivation can facilitate ICC. Methodology To acquire the participants’ views of IQ, we designed a survey as a gamified process. Additionally, we selected Web study to reach a broader audience. We increased the validity of the research by including a diverse set of participants (i.e. individuals with different education, demographic and social backgrounds). Findings The study results indicate that motivation improved the ICC of IQ on average by 0.27, demonstrating an increase in measurability from poor (0.29) to moderate (0.56). The results reveal a positive correlation between motivation level and ICC, with a significant overall increase in ICC relative to previous studies. The research also identified trends in ICC for different dimensions of IQ with the best results achieved for completeness and accuracy. Practical implications The work has important practical implications for future IQ research and suggests valuable guidelines. The results of this study imply that considering raters’ motivation improves the measurability of IQ substantially. Originality Previous studies addressed ICC in IQ dimension evaluation. However, assessors’ motivation has been neglected. This study investigates the impact of assessors’ motivation on the measurability of IQ. Compared to the results in related work, the level of agreement achieved with the most motivated group of participants was superior.

Ultrasound is widely used as a noninvasive method in therapeutic and diagnostic applications. These can be further optimized by computational approaches, as they allow for controlled testing and rational optimization of the ultrasound parameters, such as frequency and amplitude. Usually, continuum numerical methods are used to simulate ultrasound propagating through different tissue types. In contrast, ultrasound simulations using particle description are less common, as the implementation is challenging. In this work, a dissipative particle dynamics model is used to perform ultrasound simulations in liquid water. The effects of frequency and thermostat parameters are studied and discussed. We show that frequency and thermostat parameters affect not only the attenuation but also the computed speed of sound. The present study paves the way for development and optimization of a virtual ultrasound machine for large-scale biomolecular simulations.

We present a novel smart visibility system for visualizing crowded volumetric data containing many object instances. The presented approach allows users to form groups of objects through membership predicates and to individually control the visibility of the instances in each group. Unlike previous smart visibility approaches, our approach controls the visibility on a per-instance basis and decides which instances are displayed or hidden based on the membership predicates and the current view. Thus, cluttered and dense volumes that are notoriously difficult to explore effectively are automatically sparsified so that the essential information is extracted and presented to the user. The proposed system is generic and can be easily integrated into existing volume rendering applications and applied to many different domains. We demonstrate the use of the volume conductor for visualizing fiber-reinforced polymers and intracellular organelle structures.

Virtual Reality (VR) is increasingly used for visiting historic places. Research on VR experiences in dark tourism (that focuses on mortality) focuses almost exclusively on adults. No studies were found that used virtual tours to engage children with their own country’s conflicts. The present study addresses this gap by designing and developing virtual tours in four cities of Europe with a troubled past. Virtual tours engage children and youth in historical conflicts using multi-perspective storytelling. The aim of this pre-test post-test comparative case study is to examine the change on students’ perceptions of their country’s troubled past after their interaction with a virtual tour of their capital. A secondary aim is to document students’ evaluation of the virtual tour. A questionnaire examining students’ perceptions was completed before and after students’ individual interaction with a virtual tour. Participants included 360 students (212 from Cyprus, 42 from Germany, 63 from Bosnia-Herzegovina, and 44 from Kosovo). Findings indicate a statistically significant positive change in perceptions of troubled pasts for primary/secondary students from Cyprus, Bosnia-Herzegovina, and Germany. Preliminary results are promising and indicate the effectiveness of virtual tours as tools that can have an effect on students’ perceptions of troubled pasts, particularly for children rather than young adults. Students’ evaluation of the virtual tours was positive, irrespectively of participants’ age, indicating high acceptability. Bibliografija: str. 179-181. Abstract.

In today’s times, when digitalisation is advancing in all areas, travel agencies need to follow trends to optimise their work as much as possible and stay competitive. Business information systems provide a more detailed insight into business processes and their operation, which helps monitor progress and adapt to possible changes to achieve the best results in the business world. Travel agencies need a quality business information system that helps employees shift from using basic office tools to more modern and effective systems for reporting and monitoring work performance. Data is the basis on which agencies adjust, optimise and build their business. Data is available in huge quantities and a variety of formats. Raw data that is not organised into a meaningful whole cannot bring added value to the company. The thesis aims to use Power BI as a tool for editing and integrating databases in travel agencies, as well as their visualisation and display, which is understandable for the user. With the help of the tool, we will create a report which will help the travel agency improve its business and make critical decisions that are critical for success. V današnjih časih, ko digitalizacija na vseh področjih napreduje, morajo turistične agencije slediti trendom da lahko čim bolj optimizirajo svoje poslovanje in tako ostanejo konkurenčne. Poslovno-informacijski sistemi omogočajo podrobnejši vpogled v poslovne procese in njihovo delovanje, kar pomaga sledenju napredka in prilagajanju morebitnim spremembam, da bi lahko dosegli čim boljše rezultate v poslovnem svetu. Turistične agencije potrebujejo kvaliteten poslovni informacijski sistem, ki zaposlenim pomaga pri prehodu z osnovnih pisarniških orodij na sodobnejše in učinkovitejše sisteme za poročanje in spremljanje delovne uspešnosti. Podatki so osnova na kateri agencije prilagajajo, optimizirajo in gradijo svoj posel. Na voljo so v ogromnih količinah in v različnih oblikah. Neobdelani podatki, ki niso organizirani v neko smisleno celoto, ne morejo prinašati dodane vrednosti podjetju. Cilj diplomske naloge je uporabiti Power BI kot orodje za urejanje in integracijo podatkovnih baz v turistični agenciji, ter njihovo vizualizacijo in prikaz, ki je razumljiv za uporabnika. S pomočjo orodja bomo ustvarili poročilo kar bo pomagalo turustični agenciji izboljšati svoj posel in sprejemati pomembne odločitve, ki so bistvene za uspeh.

Oskrba Slovenije z električno energijo je tesno povezana z vzdrževanjem vseh obstoječih prenosnih objektov. V delu analiziramo objekt RTP Divača, čigar zelo pomembna gradnika sta prečna transformatorja s stikaloma, ki kontrolirata regulacijo napetosti in moči. Pregledi teh dveh transformatorjev se izvajajo glede na čas ali pa glede na stanje naprave. Pregledi so povezani s številom opravljenih preklopov stikal. Cilj diplomskega dela je boljše načrtovanje pregledov in vzdrževanje naprav, za kar uporabimo strojno učenje za gradnjo modelov, ki napovedujejo število dnevnih, tedenskih ter meseščnih preklopov. Modeli so se učili na podlagi podatkov izmerjenih v časovnem obdobju od leta 2017 do leta 2019, njihove napake pa so bile vrednotene na podatkih pridobljenih v časovnem obdobju od leta 2020 do leta 2021. Rezultati vrednotenja kažejo, da je za dnevno napovedovanje števila preklopov najuspešnejši model LSTM, za tedensko in mesečno napovedovanje števila preklopov pa je najuspešnejši model linearne regresije. The supply of electricity to Slovenia is closely connected with the maintenance of all existing portable facilities in Slovenia. In this B.A. thesis, we analyse the RTP Divača facility, whose very important building blocks are two phase-shifting transformer with taps, which control voltage and power regulation. The checks of these two transformers are carried out in regard to time or the state of the transformer. In the thesis, the number of switches per day, per week and per month is modelled with the help of machine learning. The models were trained based on the data measured in the time period from 2017 to 2019, and their errors were evaluated on the basis of the data measured in the period from 2020 to 2021. The results of the test data show that for daily forecasting the number of switches LSTM is the most successful model, while for weekly and monthly forecasting the number of switches a linear regression model is best.

V delu definiramo verjetnostne kontekstno-neodvisne gramatike in opišemo njihovo uporabo v simbolni regresiji. Omejimo se na gramatike, ki tvorijo algebrajske izraze in natančno definiramo izraz. Motiviramo problem izračuna verjetnosti podanega izraza za podano gramatiko in dokažemo, da je v splošnem neodločljiv. Predstavimo nekaj posebnih primerov gramatik, ki generirajo izraze, za katere je problem algoritmično rešljiv. Podamo postopek za izračun verjetnosti podanega izraza za te posebne primere gramatik. Na koncu predstavimo verjetnostne gramatike, ki generirajo podmnožice grup. We define probability context-free grammars and describe their use for symbolic regression. We study grammars that generate algebraic expressions and meticulously define an expression. We show that calculating the probability of a given expression for a given grammar is generally undecidable. We overview specific grammars for generating expressions, where an algorithm for calculating the probability of a given expression exists. For those example grammars, we present an algorithm for calculating the probability of a given expression. At last, we present probabilistic grammars for generating elements from groups.

V delu predstavimo algoritmično trgovanje. Gre za trgovanje, kjer je trgovalna strategija, ki je zaporedje pravil, ki določajo kaj, kdaj in koliko nekega sredstva kupimo oziroma prodamo, opisljiva z algoritmi.Predstavljeni so tudi avtomatski trgovalni sistemi, ki za avtomatizacijo potrebujejo algoritmično trgovanje. Navadno je algoritmično trgovalna strategija narejena na podlagi temeljne in tehnične analize, zato definiramo različne tehnične indikatorje s katerimi to naredimo. Poleg slednjih dveh pa lahko strategijo sestavimo tudi z metodami strojnega učenja s katerimi naredimo napovedni model. V ta namen definiramo nekaj glavnih metod in vse potrebne funkcije, da lahko uspešnost modela izračunamo. V nadaljevanju potem naredimo napovedni model za trgovanje s kriptovalutami. Podatke izboljšamo s tem, ko dodamo še tehnične indikatorje kot napovedne spremenljivke. Na teh podatkih potem naučimo modele, ki smo jih spoznali in jih med seboj primerjamo. In the thesis we introduce the concept of algorithmic trading, which is the type of trading that uses algorithms to execute trading strategies. Strategy is a set of rules that define what, when and how much of an asset we should trade. Furthermore, we give an overview of the automated trading system, which requires algorithmic trading for automation. Algorithmic strategy is usually built on the basis of fundamental and technical analysis, which is why we define several technical indicators, that are used in trading algorithms. Moreover it can be constructed with machine learning methods. We present several of them as well as all the necessary functions to evaluate prediction models. Then we follow up with the practical part, where we create our own trading strategy based on the machine learning methods we covered, using data enhanced with technical indicators. Finally we evaluate the models and present the results.

V prvem delu diplomskega dela je predstavljena platforma Ethereum. Ethereum je decentralizirana računalniška infrastruktura, namenjena zaganjanju programov, imenovanih pametne pogodbe. Izpostavljene in razložene so glavne komponente omrežja oz. bločne verige. Prav tako je opisano kriptografsko ozadje delovanja omrežja. Bolj podrobno je razložen programski jezik za pisanje pogodb, Solidity. Orisane so aplikacije, ki za svoje delovanje uporabljajo pametne pogodbe, imenovane decentralizirane aplikacije oz. dApps. Predstavljeni so oraklji, to so sistemi, ki v bločno verigo vnesejo podatke izven verige, in izpostavljen je problem decentraliziranosti orakljev. Opisana in matematično formulirana je rešitev omrežja Flare, orakelj FTSO. Drugi del naloge opisuje razvoj pametne pogodbe SimpleOracle ter izpostavlja bistvene dele pogodbe. Priložena so tudi navodila za vzpostavitev delovnega okolja za pisanje pametnih pogodb ter razložena uporaba nekaterih orodij. The first part of the thesis describes the Ethereum platform. Ethereum is a decentralized computer infrastructure for executing programs called smart contracts. The main components of the blockchain are outlined and characterized. There is also cryptographic background of the network explained. The programming language for smart contracts, Solidity, is explained in a bit more detailed manner. Decentralized applications (applications that use smart contracts technology) are described. Systems that provide the blockchain with data from the outside of the chain, oracles, are presented and the decentralized oracle problem is defined. Flare network's FTSO oracle is presented together with relevant mathematical formulations. The second part of the thesis details the implementation of smart contract SimpleOracle and outlines the main parts of the contract. Instructions for setting up the work environment are provided, and the use of some tools is demonstrated.

Predmet raziskovanja tega magistrskega dela je problem zapiranja centrov, ki spada med probleme razmeščanja ponudnikov. Predstavljene so osnovne lastnosti ter natančni algoritmi za reševanje problema. Zaradi NP-polnosti so predstavljeni tudi aproksimacijski algoritmi. Empirični testi na različnih omrežjih preverjajo časovno zahtevnost in kakovost rešitev aproksimacijskih algoritmov. Razmerje teh dveh lastnosti je prav tako primerjano. V namen testiranja algoritmov je razvito tudi poštno omrežje Slovenije, cilj katerega je reševanje problema zapiranja poštnih poslovalnic. The main topic of this thesis is the center closing problem, which is a facility location problem. Basic properties and exact algorithms for solving the problem are presented. Because the problem is NP-complete, approximation algorithms are also presented. Time complexity, the quality of results, and the ratio between the two are tested for approximation algorithms using empirical tests. For the purpose of testing the algorithms a postal network of Slovenia was constructed, the goal of which was to solve the post office closing problem.

Iskanje relacij med entitetami v besedilu je področje obdelave naravnega jezika. Pri iskanju relacij želimo v stavku: "Ljubljana je glavno mesto Slovenije" odkriti, da med entitetama Ljubljana in Slovenija nastopa relacija glavno mesto. V zaključnem delu smo najprej naredili pregled metod za učenje modelov za napovedovanje relacij. Nato smo si izbrali tri metode z različnimi pristopi za učenje modelov, ki napovedujejo relacije. Metodo s povratno nevronsko mrežo z dolgim kratkoročnim spominom, metodo z vložitvami BERT in metodo RECON, ki uporabi grafovsko nevronsko mrežo s pozornostjo. Za učenje modelov smo uporabili slovenski korpus, ki smo ga polavtomatsko generirali iz besedil slovenske Wikipedije. Naučene modele smo nato testirali na testnem korpusu besedil slovenske Wikipedije in testnem korpusu člankov strani 24ur.com. Na testnem korpusu slovenske Wikipedije so vse tri metode dosegle visoke priklice in točnosti, najbolje se je odrezala metoda RECON. Veliko slabše rezultate so dosegle na testni množici člankov 24ur.com, kjer se je še najbolje izkazala metoda z vložitvami BERT, ko je uporabila vložitve CroSloEngual. Finding relations between entities in a text is an area of natural language processing. In the sentence: "Ljubljana is the capital of Slovenia" we want to find the relation capital between entities Ljubljana and Slovenia. We first start with a review of the methods used for training models to predict relations. We then chose three methods with different approaches. The method with long short-term memory neural network, method which uses BERT encoder representations and method RECON which uses graph attention networks. To train the models, we used the Slovenian corpus which was generated semi-automatically from the text of the Slovenian Wikipedia. We test the models on a test corpus of Slovenian Wikipedia and the test corpus of articles on 24ur.com. All three methods achieved high recall and precision for the test corpus of the Slovenian Wikipedia and the RECON method performed best. Results were worse on the test set of 24ur.com articles, where the method which used BERT encoder representations CroSloEngual achieved the best results.

Untimely interaction in critical moments, content that does not capture a user’s attention, and missed opportunities for delivering relevant information are just some of the issues caused by ubiquitous computing devices’ inability to understand their users cognitive engagement. However, current means of inferring cognitive load are, with a few exceptions, based on intrusive methods requiring physical contact of the measurement equipment and the user, thus of limited applicability in ubiquitous computing. In this paper we propose Wi-Mind, a system for remote cognitive load assessment through wireless sensing. Wi-Mind is based on a software-defined radio radar that measures sub-millimeter movements related to a person’s breathing and heart beats, which, in turn allows us to infer the person’s cognitive load. We built and tested the system with 23 volunteers engaged in different tasks. Initial results show that while Wi-Mind manages to detect whether one is engaged in a cognitively demanding task, the inference of the exact cognitive load level remains challenging.

Edge intelligence is currently facing several important challenges hindering its performance, with the major drawback being meeting the high resource requirements of deep learning by the resource-constrained edge computing devices. The most recent adaptive neural network compression techniques demonstrated, in theory, the potential to facilitate the flexible deployment of deep learning models in real-world applications. However, their actual suitability and performance in ubiquitous or edge computing applications has not, to this date, been evaluated. In this context, our work aims to bridge the gap between the theoretical resource savings promised by such approaches and the requirements of a real-world mobile application by introducing algorithms that dynamically guide the compression rate of a neural network according to the continuously changing context in which the mobile computation is taking place. Through an in-depth trace-based investigation, we confirm the feasibility of our adaptation algorithms in offering a scalable trade-off between the inference accuracy and resource usage. We then implement our approach on real-world edge devices and, through a human activity recognition application, confirm that it offers efficient neural network compression adaptation in highly dynamic environments. The results of our experiment with 21 participants show that, compared to using static network compression, our approach uses 2.18× less energy with only a 1.5% drop in the average accuracy of the classification.

Self-regulated learning (SRL) plays an important role in successful learning with hypertexts. The use of appropriate SRL strategies helps students acquire new knowledge more efficiently. We investigated the use of SRL strategies in individual learning from expository science hypertext, the correlations between different measures of self-regulation, and the differences in SRL between more and less successful students. A sample of 443 ninth graders from 15 different schools participated in the study. A variety of off- and online measures were used to measure SRL. Data were collected from student traces, questionnaires, tests scores, and notes. Low correlations between the off- and online measures of SRL suggest that they measure different aspects of SRL use in learning digital science texts. Student achievement in science positively correlated with their use of SRL strategies. Students with higher knowledge gains reported the higher use of deep cognitive strategies, higher motivation for learning, and used a higher number of strategies in note-taking while learning. The results of this study may have practical implications for teachers to support student SRL and for developers of digital learning materials to incorporate SRL scaffolding into learning topics.

Nasl. z nasl. zaslona. Opis vira z dne 4. 4. 2022. Bibliografija: str. 21-22. Abstract.

Nowadays, recommender systems are vital in lessening the information overload by filtering out unnecessary information, thus increasing comfort and quality of life. Matrix factorization (MF) is a well-known recommender system algorithm that offers good results but requires a certain level of system knowledge and some effort on part of the user before use. In this article, we proposed an improvement using grammatical evolution (GE) to automatically initialize and optimize the algorithm and some of its settings. This enables the algorithm to produce optimal results without requiring any prior or in-depth knowledge, thus making it possible for an average user to use the system without going through a lengthy initialization phase. We tested the approach on several well-known datasets. We found our results to be comparable to those of others while requiring a lot less set-up. Finally, we also found out that our approach can detect the occurrence of over-saturation in large datasets.

Magistrsko delo obravnava tehnološke zmožnosti, ki nam jih ponujajo sodobne programske rešitve pri implementaciji aplikacij, ki so nameščene na pametnih televizorjih. V delu je najprej predstavljena zgodovina multimedijskih vsebin na televizorjih, poudarek je predvsem na tehnologiji teletekst, katerega osnovno delovanje je opisano. Predstavljene so faze prehoda TV naprav in oddajnikov iz analognih v digitalne sisteme, ki so odprli nove načine razvoja aplikacij in storitev za končne uporabnike. Obravnavi standarda HbbTV sledi osrednji del dela, kjer je obravnavan potek načrtovanja aplikacij in razvojnih vzorcev, ki predstavljajo sestavni del razvoja. Prikazani so praktični primeri različnih implementacij, tako aplikacij, ki ponujajo predvajanje na oddaljenem zaslonu, kot tudi aplikacije, ki s pomočjo tehnologije »WebView« deluje na Android TV sistemih. Za slednjo sta prikazana načrt in struktura, ki omogočata delovanje. Prenovljena aplikacija je bila izdelana z uporabo metode RAD (ang. Rapid Application Development), ki je primerna za manjše ekipe in je osnovana na prototipih, saj pri samem razvoju niso pričakovane bistvene spremembe v programju. V zadnjem delu je obravnavan postopek objave aplikacije v »spletni trgovini«. The master’s thesis deals with technological capabilities offered by modern software solutions when implementing apps, which are installed on smart televisions. At the beginning of the thesis is laid out the history of multimedia content on television, with a particular emphasis on the teletext technology, the basic functioning of which is described. The transition from TV devices and analogue transmitters to digital systems, which introduced new ways of developing apps and services for the users, is presented. In the next chapter the HbbTV standard is addressed, followed by the core of the thesis where the course of app development and development patterns, which are a constituent part of development, is discussed. The work provides practical examples of different implementations of apps, which offer »viewing for remote screens«, as well as the implementation of an app which operates on Android TV system with the help of »WebView« technology. A plan and structure, which enable its functioning, are displayed for the latter. The app has been developed with the RAD (ang. Rapid Application Development) methodology which is suitable for smaller teams and is based on prototypes, as no significant software changes are expected in the development phase. In the last part of the thesis the procedure of release of the app on »online shop« is discussed.

Blockchain ecosystems are rapidly maturing and meeting the needs of business environments (e.g., industry, manufacturing, and robotics). The decentralized approaches in industries enable novel business concepts, such as machine autonomy and servitization of manufacturing environments. Introducing the distributed ledger technology principles into the machine sharing and servitization economy faces several challenges, and the integration opens new interesting research questions. Our research focuses on data and event models and secure upgradeable smart contract platforms for machine servitization. Our research indicates that with the proposed approaches, we can efficiently separate on- and off-chain data and assure scalability of the DApp without compromising the trust. We demonstrate that the secure upgradeable smart contract platform, which was adapted for machine servitization, supports the business workflow and, at the same time, assures common identification and authorization of all the participants in the system, including people, devices, and legal entities. We present a hybrid decentralized application (DApp) for the servitization of 3D printing. The solution can be used for or easily adapted to other manufacturing domains. It comprises a modular, upgradeable smart contract platform and off-chain machine, customer and web management, and monitoring interfaces. We pay special attention to the data and event models during the design, which are fundamental for the hybrid data storage and DApp architecture and the responsiveness of off-chain interfaces. The smart contract platform uses a proxy contract to control the access of smart contracts and role-based access control in function calls for blockchain users. We deploy and evaluate the DApp in a consortium blockchain network for performance and privacy. All the actors in the solution, including the machines, are identified by their blockchain accounts and are compeers. Our solution thus facilitates integration with the traditional information-communication systems in terms of the hybrid architectures and security standards for smart contract design comparable to those in traditional software engineering.

Raglio, A., Baiardi, P., Vizzari, G., Imbriani, M., Castelli, M., Manzoni, S., Vico, F., & Manzoni, L. (2021). Algorithmic music for therapy: Effectiveness and perspectives. Applied Sciences (Switzerland), 11(19), 1-13. [8833]. https://doi.org/10.3390/app11198833 --------------------------------------------------------------Funding Information: Acknowledgments: This work was partially supported by FCT, Portugal, through funding of the project GADgET (DSAIPA/DS/0022/2018), and the financial support from the Slovenian Research Agency (research core funding no. P5-0410). Publisher Copyright: © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This study assessed the short-term effects of conventional (i.e., human-composed) and algorithmic music on the relaxation level. It also investigated whether algorithmic compositions are perceived as music and are distinguishable from human-composed music. Three hundred twenty healthy volunteers were recruited and randomly allocated to two groups where they listened to either their preferred music or algorithmic music. Another 179 healthy subjects were allocated to four listening groups that respectively listened to: music composed and performed by a human, music composed by a human and performed by a machine; music composed by a machine and performed by a human, music composed and performed by a machine. In the first experiment, participants underwent one of the two music listening conditions—preferred or algorithmic music— in a comfortable state. In the second one, participants were asked to evaluate, through an online questionnaire, the musical excerpts they listened to. The Visual Analogue Scale was used to evaluate their relaxation levels before and after the music listening experience. Other outcomes were evaluated through the responses to the questionnaire. The relaxation level obtained with the music created by the algorithms is comparable to the one achieved with preferred music. Statistical analysis shows that the relaxation level is not affected by the composer, the performer, or the existence of musical training. On the other hand, the perceived effect is related to the performer. Finally, music composed by an algorithm and performed by a human is not distinguishable from that composed by a human. publishersversion published

Living Labs are spaces for innovative and participative research, development and activities that use multidisciplinary approaches and promote the co-creation paradigm. Our specific interest lies in exploring the value of the Living Lab concept for creating environments that enable equal opportunities for people living in rural and urban areas, and for making rural areas attractive places to live. Moreover, through the existing practices and research results available, Living Labs are seen as one of the important building blocks of smart rural development and an important step towards establishing a Smart Village environment. Living Labs are a valuable player in enhancing circular economy, digital transformation, local self-sufficiency and other elements of sustainable living. The main aim of this paper is therefore to put Living Labs in the context of rural areas and evaluate their possible contributions for sustainable rural development. This paper argues that the element of community and social change should be considered as a key element in enabling sustainable living.

Recently, a derivative-free optimization algorithm was proposed that utilizes a minimum Frobenius norm (MFN) Hessian update for estimating the second derivative information, which in turn is used for accelerating the search. The proposed update formula relies only on computed function values and is a closed-form expression for a special case of a more general approach first published by Powell. This paper analyzes the convergence of the proposed update formula under the assumption that the points from Rn where the function value is known are random. The analysis assumes that the N+2 points used by the update formula are obtained by adding N+1 vectors to a central point. The vectors are obtained by transforming a prototype set of N+1 vectors with a random orthogonal matrix from the Haar measure. The prototype set must positively span a N≤n dimensional subspace. Because the update is random by nature we can estimate a lower bound on the expected improvement of the approximate Hessian. This lower bound was derived for a special case of the proposed update by Leventhal and Lewis. We generalize their result and show that the amount of improvement greatly depends on N as well as the choice of the vectors in the prototype set. The obtained result is then used for analyzing the performance of the update based on various commonly used prototype sets. One of the results obtained by this analysis states that a regular n-simplex is a bad choice for a prototype set because it does not guarantee any improvement of the approximate Hessian.

Although proposed more than half a century ago, the Nelder–Mead simplex search algorithm is still widely used. Four numeric constants define the operations and behavior of the algorithm. The algorithm with the original constant values performs fine on most low-dimensional, but poorly on high-dimensional, problems. Therefore, to improve its behavior in high dimensions, several adaptive schemas setting the constants according to the problem dimension were proposed in the past. In this work, we present a novel adaptive schema obtained by a meta-optimization procedure. We describe a schema candidate with eight parameters subject to meta-optimization and define an objective function evaluating the candidate’s performance. The schema is optimized on up to 100-dimensional problems using the Parallel Simulated Annealing with Differential Evolution global method. The obtained global minimum represents the proposed schema. We compare the performance of the optimized schema with the existing adaptive schemas. The data profiles on the Gao–Han modified quadratic, Moré–Garbow–Hilstrom, and CUTEr (Constrained and Unconstrained Testing Environment, revisited) benchmark problem sets show that the obtained schema outperforms the existing adaptive schemas in terms of accuracy and convergence speed.

Edge computing brings artificial intelligence algorithms and graphics processing units closer to data sources, making autonomy and energy-efficient processing vital for their design. Approximate computing has emerged as a popular strategy for energy-efficient circuit design, where the challenge is to achieve the best tradeoff between design efficiency and accuracy. The essential operation in artificial intelligence algorithms is the general matrix multiplication (GEMM) operation comprised of matrix multiplication and accumulation. This paper presents an approximate general matrix multiplication (AGEMM) unit that employs approximate multipliers to perform matrix–matrix operations on four-by-four matrices given in sixteen-bit signed fixed-point format. The synthesis of the proposed AGEMM unit to the 45 nm Nangate Open Cell Library revealed that it consumed only up to 36% of the area and 25% of the energy required by the exact general matrix multiplication unit. The AGEMM unit is ideally suited to convolutional neural networks, which can adapt to the error induced in the computation. We evaluated the AGEMM units’ usability for honeybee detection with the YOLOv4-tiny convolutional neural network. The results implied that we can deploy the AGEMM units in convolutional neural networks without noticeable performance degradation. Moreover, the AGEMM unit’s employment can lead to more area- and energy-efficient convolutional neural network processing, which in turn could prolong sensors’ and edge nodes’ autonomy.

V današnjem času se v industriji pri tehnoloških procesih generira ogromna količina podatkov, ki so neuporabni, če se jih ne analizira in pridobi znanje iz teh podatkov na pravi način in v primernem času. Zato v teh primerih tradicionalne statistične metode včasih niso dovolj informativne, ker je treba vnaprej definirati hipotezo, ki jo kasneje potrdimo ali zavržemo. S podatkovnim rudarjenjem to ni potrebno, saj dobimo modele iz podatkov na avtomatičen način, ki lahko predstavlja novo znanje in iz tega lahko definiramo neko hipotezo. Industrijski in avtomobilski trg rasteta iz dneva v dan in da bi sledili razvoju ter analizi nastalih podatkov, moramo izbrati potrebne ukrepe, ki bi nam olajšali samo analiziranje prejetih informacij in ustrezno reagirali ob možnih odstopanjih, da bi s tem zmanjšali potencialne nastale stroške kot samo izvedbo procesa. Na eni izmed linij za montažo enega primera motorja v podjetju MAHLE Electric Drives Slovenija, d. o. o., smo iz operacij procesa vtiska ležaja v pokrov, zakovanja ležaja, natiska ležaja na rotor in vtiska rotorja videli, da kljub temu, da so operacije pri meritvah sile in poti še vedno v definiranih tolerancah, imamo precej odstopanj od referenčne vrednosti in s tem je lahko povezan tudi rezultat meritve trenja motorja. S pomočjo podatkovnega rudarjenja bi radi ugotovili, kaj je razlog za ta odstopanja, da bi jih lahko vnaprej napovedali in sprejeli ustrezne ukrepe. Za lažje razumevanje omenjene tematike in kot pomoč tehnologom na liniji pri pregledovanju potencialnih odstopanj in napovedovanju ustreznosti motorja smo pripravili tudi programski vmesnik. Nowadays, technological industrial processes generate an enormous amount of data, which can become useless if they are not analysed and relevant information is not extracted appropriately and within a reasonable period of time. In many cases, traditional statistical methods do not provide enough information, because hypotheses need to be defined in advance, and later confirmed or refuted. However, this is not necessary when the data mining method is applied, because data are acquired from models automatically. The data may convey new information, which can be used to set a hypothesis. The industrial car market is growing day by day in order to follow its development and data analysis, several methods must be selected to facilitate data analysis and help us react appropriately in the event of probable deviations in order to reduce potential costs and shorten the process itself. At the MAHLE Electric Drives Slovenija company, we worked on one of the assembly lines for engine installation, where bearings get embedded in the lid, forged and embedded onto the rotor, which also gets embedded. Even though these operations are within the defined tolerances when force and stroke are measured, several deviations from the referential value are identified, which may influence the result of engine friction measurement. Using the data mining method, we want to determine the reason behind these deviations in order to be able to foresee them and adopt appropriate measures. Seeking to facilitate the understanding of the discussed topic as well as to help engineers on the line review potential deviations and predict the quality of the engine, we also created a software interface. Findings obtained with the help of data mining techniques will be very useful in planning new lines and optimizing production processes before the regular start of production, according to the data obtained, we can use the software interface to see the effects on the final measurements and we can predict the suitability of the engine itself.

Over the last decade, the advent of the cluster ensemble framework has enabled more accurate and robust data analysis than traditional single clustering algorithms. The improved clustering of microarray data has had a particularly strong impact in the fields of genomics and medicine. However, when we bring several ensemble members together to form a consensus, low-quality data partitions can seriously compromise the final solution. One way to overcome this problem is the weighted cluster ensemble approach based on Partition Relevance Analysis (PRA), which uses internal cluster validity indices to evaluate and weight the ensemble members before the fusion. Unfortunately, the selection of appropriate validation indices for given data is far from trivial. In this paper, we propose an additional step in PRA that reduces the size of the committee of cluster validation indices. It does so by eliminating redundant and noisy indices using data dimensionality reduction methods. Our extension works in an unsupervised way, minimizing the amount of user intervention and required expert knowledge. We adapted three conventional consensus functions based on the principle of evidence accumulation to work with PRA weights. We demonstrate the advantages of the proposed reduction step of PRA based on extensive experiments with 25 gene expression and 15 non-genetic real-world datasets, where we compared 15 consensus functions. The source code is available at https://github.com/nejci/PRAr.

Kernel methods provide a principled way for general data representations. Multiple kernel learning and kernel approximation are often treated as separate tasks, with considerable savings in time and memory expected if the two are performed simultaneously. Our proposed Mklaren algorithm selectively approximates multiple kernel matrices in regression. It uses Incomplete Cholesky Decomposition and Least-angle regression (LAR) to select basis functions, achieving linear complexity both in the number of data points and kernels. Since it approximates kernel matrices rather than functions, it allows to combine an arbitrary set of kernels. Compared to single kernel-based approximations, it selectively approximates different kernels in different regions of the input spaces. The LAR criterion provides a robust selection of inducing points in noisy settings, and an accurate modelling of regression functions in continuous and discrete input spaces. Among general kernel matrix decompositions, Mklaren achieves minimal approximation rank required for performance comparable to using the exact kernel matrix, at a cost lower than 1% of required operations. Finally, we demonstrate the scalability and interpretability in settings with millions of data points and thousands of kernels.

Nasl. z nasl. zaslona. Opis vira z dne 26. 7. 2021. Bibliografija: str. 86. Abstract.

This paper presents the results of a user study of the effects of different head-centric rest-frames on Virtual Reality-Induced Symptoms and Effects (VRISE) and the user experience in virtual environments (VE). Participants played the custom-designed 3D game in two different game modes (high action and low action). For assessing VRISE levels, we used the Simulator Sickness Questionnaire (SSQ) and Fast Motion Sickness Score (FMS). The presence was evaluated by SPES (Spatial Presence Experience Scale), and for the user experience, the short version of the User Experience Questionnaire (UEQ-S) was used. The results indicate that the usage of head-centric rest-frames negatively affected VRISE levels (more sickness) in the low action mode of the game. However, for the users experienced with VR technology, the VRISE disorientation symptoms were alleviated in a high action mode of the game with rest-frame glasses. We found no negative effect of rest-frames on the user experience and presence, except for some negative impact when using rest-frame glasses in the low action mode of the game. No negative impact on the performance itself was observed. That means that the usage of head-centric rest-frames is suitable for usage in VR applications. In terms of VRISE levels, we found out that rest-frame glasses are more suitable for the wearers of the distance spectacles, and a baseball hat is more suitable for non-wearers of distance spectacles.

This paper presents a new A-weighting filter&rsquo, s design and explores the potential of using approximate multiplication for low-power digital A-weighting filter implementation. It presents a thorough analysis of the effects of approximate multiplication, coefficient quantization, the order of first-order sections in the filter&rsquo, s cascade, and zero-pole pairings on the frequency response of the digital A-weighting filter. The proposed A-weighting filter was implemented as a sixth-order IIR filter using approximate odd radix-4 multipliers. The proposed filter was synthesized (Verilog to GDS) using the Nangate45 cell library, and MATLAB simulations were performed to verify the designed filter&rsquo, s magnitude response and performance. Synthesis results indicate that the proposed design achieves nearly 70% reduction in energy (power-delay product) with a negligible deviation of the frequency response from the floating-point implementation. Experiments on acoustic noise suggest that the proposed digital A-weighting filter can be deployed in environmental noise measurement applications without any notable performance degradation.

Collecting complex information on the status of machinery is the enabler for advanced maintenance activities, and one of the main players in this process is the sensor. This paper describes modern maintenance strategies that lead to Condition-Based Maintenance. This paper discusses the sensors that can be used to support maintenance, as of different categories, spanning from common off-the-shelf sensors, to specialized sensors monitoring very specific characteristics, and to virtual sensors. This paper also presents four different real-world examples of project pilots that make use of the described sensors and draws a comparison between them. In particular, each scenario has unique characteristics requiring different families of sensors, but on the other hand provides similar characteristics on other aspects., This work was partially supported by National Funds through FCT/MEC (Portuguese Foundation for Science and Technology), Slovenian Research Agency (research core funding No. P2-0098), Finnish Funding Agency for Innovation Tekes, and co-financed by ERDF (European Regional Development Fund) under the PT2020 Partnership, within the CISTER Research Unit (PCEC/04234); also by the EU ECSELJU under the H2020 Framework Programme, within project ECSEL/0004/2014, JU grant no. 662189 (MANTIS); also by EU ECSEL JU under the H2020 Framework Programme, JU grant no. 737459 (Productive4.0 project).

Množenje matrik je v linearni algebri preprosta operacija, ki se pogosto pojavlja v rešitvah najrazličnejših problemov. Prav zato je bilo v iskanje hitrih algoritmov za množenje matrik vloženega že veliko dela. V diplomskem delu definiramo problem iskanja zgornje meje eksponenta matričnega množenja in razvijemo teorijo ranga in mejnega ranga bilinearnih preslikav. Predstavimo več algoritmov za hitro množenje matrik, ki slonijo na tej teoriji. Izbrane algoritme tudi implementiramo, jih primerjamo med seboj in ocenimo njihovo uporabnost v praksi. Matrix multiplication is one of the most basic operations in linear algebra and thus very common in various scientific disciplines. Consequently, the computation complexity of matrix multiplication has been extensively studied. In this work, we define a problem of finding the upper bound for the exponent of matrix multiplication and present the theory of rank and border rank of bilinear maps. We describe multiple fast matrix multiplication algorithms based on this theory. In the end, we implement some selected algorithms, compare them, and discuss their value in practical applications.

Namen prispevka: Mesta postajajo tehnološko naprednejša, zlasti, da bi zadostila potrebam vedno večjega števila ljudi. Tehnologija, predvsem »pametna« tehnologija, daje posamezniku udobje v zameno za njegovo zasebnost. Dostop do storitev je mogoč s pomočjo številnih programskih rešitev, tudi aplikacij na mobilnih napravah. Ob nevestni uporabi aplikacij ter nepoznavanju pomena varovanja zasebnosti je tveganje za poseg v zasebnost veliko. Namen prispevka je izpostaviti stališče ljudi do zasebnosti v pametnih mestih ter predstaviti, kaj so tehnologije, ki sestavljajo pametno mesto. Metode: Predstavljene ugotovitve so podprte z deskriptivnimi dognanji, ki temeljijo na virih in literaturi ter izvedeni raziskavi, ki smo jo analizirali s pomočjo statističnih metod. Ugotovitve: Poznavanje koncepta pametnih mest je na slovenskem področju izjemno slabo. Glavne ugotovitve raziskave kažejo, da ljudje niso pripravljeni bivati v pametnih mestih. Skrbi jih nivo potrebnega računalniškega znanja, ki izhaja iz posameznikovega nepoznavanja pametnih mest. Podatke, pridobljene v pametnih mestih, bi morala upravljati država in ne zasebna podjetja. Omejitve/uporabnost raziskave Znanstvenih objav na temo zasebnosti v pametnih mestih je malo. Omejitev predstavlja tudi ciljna skupina, ki ni na družabnih omrežjih (predvsem starejša populacija), zato bi bilo smiselno izvesti raziskavo tudi med starejšo populacijo. Praktična uporabnost: Izsledki raziskave nam pokažejo posameznikovo poznavanje pomena zasebnosti v pametnih mestih, kar predstavlja izhodišče za nadaljnje aplikativno in znanstveno delo na omenjenem področju. Izvirnost/pomembnost prispevka: Prispevek na izviren način obravnava aktualno tematiko, katere pomembnost bo v prihodnosti še naraščala. Purpose: Cities are becoming technologically advanced in order to fulfil the needs of constantly increasing number of people. Namely, technology, especially smart one, provides individuals commodity in return for their privacy. Negligent use of applications and lack of knowledge on importance of safeguarding privacy induce great risk for privacy encroachment. The purpose of this paper is to emphasise the attitude of individuals toward privacy in smart cities and to present the technology of smart city. Design/Methods/Approach: The findings, presented in paper, are based on descriptive analysis of sources and literature and statistical analysis of research results. Findings: Knowledge on smart city concept in Slovenia is extremely poor. The main findings of this research show that people do not want to live in smart cities. They are worried about the needed level of computer knowledge, which even further shows, how little people know on smart cities. The data acquired in smart cities should be in the jurisdiction of the country and not in the hands of private companies. Research Limitations / Implications: There are few scientific publications on privacy in smart cities. Another limitation is the focus group of people, who do not use social networks, especially the older population. Therefore, it would be wise to conduct research among this population. Practical Implications: Findings of the research could be basis for lectures on awareness of privacy in smart cities and for guidelines for future work regarding more effective privacy protection. Originality/Value: The paper addresses topical issue, whose importance will continue to increase in the future.

Computationally efficient 3D orientation (3DO) tracking using gyroscope angular velocity measurements enables a short execution time and low energy consumption for the computing device. These are essential requirements in today&rsquo, s wearable device environments, which are characterized by limited resources and demands for high energy autonomy. We show that the computational efficiency of 3DO tracking is significantly improved by correctly interpreting each triplet of gyroscope measurements as simultaneous (using the rotation vector called the Simultaneous Orthogonal Rotation Angle, or SORA) rather than as sequential (using Euler angles) rotation. For an example rotation of 90&deg, depending on the change in the rotation axis, using Euler angles requires 35 to 78 times more measurement steps for comparable levels of accuracy, implying a higher sampling frequency and computational complexity. In general, the higher the demanded 3DO accuracy, the higher the computational advantage of using the SORA. Furthermore, we demonstrate that 12 to 14 times faster execution is achieved by adapting the SORA-based 3DO tracking to the architecture of the executing low-power ARM Cortex&reg, M0+ microcontroller using only integer arithmetic, lookup tables, and the small-angle approximation. Finally, we show that the computational efficiency is further improved by choosing the appropriate 3DO computational method. Using rotation matrices is 1.85 times faster than using rotation quaternions when 3DO calculations are performed for each measurement step. On the other hand, using rotation quaternions is 1.75 times faster when only the final 3DO result of several consecutive rotations is needed. We conclude that by adopting the presented practices, the clock frequency of a processor computing the 3DO can be significantly reduced. This substantially prolongs the energy autonomy of the device and enhances its usability in day-to-day measurement scenarios.

Direct point-cloud visualisation is a common approach for visualising large datasets of aerial terrain LiDAR scans. However, because of the limitations of the acquisition technique, such visualisations often lack the desired visual appeal and quality, mostly because certain types of objects are incomplete or entirely missing (e.g., missing water surfaces, missing building walls and missing parts of the terrain). To improve the quality of direct LiDAR point-cloud rendering, we present a point-cloud processing pipeline that uses data fusion to augment the data with additional points on water surfaces, building walls and terrain through the use of vector maps of water surfaces and building outlines. In the last step of the pipeline, we also add colour information, and calculate point normals for illumination of individual points to make the final visualisation more visually appealing. We evaluate our approach on several parts of the Slovenian LiDAR dataset.