Μεταπτυχιακές εργασίες / Master Theses

Μόνιμο URI για αυτήν τη συλλογή

https://apothesis.hmu.gr/handle/20.500.12688/9246

Περιηγούμαι

Τώρα δείχνει 1 - 20 of 525

3D view generation and view synthesis based on 2D data.
(ΕΛ.ΜΕ.ΠΑ., Σχολή Μηχανικών (ΣΜΗΧ), ΠΜΣ Πληροφορική και Πολυμέσα, 2023-10-05) Lisitsa, Evgkenia; Λισίτσα, Εβγκένια
The simplification of view synthesis and complex geometry generation is a constantly evolving topic in computer science. The need to develop an efficient methodology to recreate real-world data based on a simple dataset, such as a sequence of images, has become even more crucial with the recent technological advances that allow higher-resolution graphics even on simple devices. Extraction of 3D data from a sequence of photos (photogrammetry) is a challenging, process-intensive, and extremely time-consuming procedure. Usually, photogrammetry requires taking a great number of sample photos to efficiently extract the geometry in the form of point clouds and then to be able to post-process the geometry to generate views. The main goal of this master's thesis is to understand and examine the state of the art in photogrammetry and novel view synthesis. The focus of this thesis is to examine the creation of images directly from voxel-based data representations, postponing the intermediate step of extracting geometries. This problem appears in the literature as the "Novel View Synthesis" problem. Additionally, this thesis aims to find out the limitations that currently exist in photogrammetry and novel view synthesis approaches. By identifying these limitations, it will be possible to improve future studies and develop new techniques to reduce the user intervention and computation power required for photogrammetry.
3D visualization of learning management systems with gamification techniques to increase motivation of students.
(ΕΛ.ΜΕ.ΠΑ., Σχολή Μηχανικών (ΣΜΗΧ), ΠΜΣ Πληροφορική και Πολυμέσα, 2024-01-29) Selekos, Petros; Σελέκος, Πέτρος
This thesis presents the design and development of a 3D gamified environment for the Open e-Class Learning Management System (LMS), aiming to enhance student motivation and engagement in e-learning. E-learning, a pivotal aspect of modern education, leverages various technologies to provide accessible and flexible learning experiences. However, maintaining student interest and active participation remains a challenge in digital learning environments. To address this, the thesis explores the integration of game-based learning into LMS, capitalizing on the engaging nature of video games to create an immersive educational experience. The core concept revolves around visualizing the Open e-Class LMS in a 3D environment, where students can navigate and interact with various LMS tools through a user-friendly interface. The implementation involves two main components: the front end, developed using the Unity3D game engine, and the back end, comprising a REST API for communication with the LMS database, developed with Spring Boot framework. A significant feature of this project is the creation of dynamic environments within the game. These elements serve as interactive spaces for learners, offering a novel approach to accessing and engaging with course materials. The thesis demonstrates that by blending 3D visualization with gamification techniques like badges, leaderboards and score, it is possible to transform traditional LMS interfaces into more engaging and interactive learning platforms. Additionally, transforming traditional exercises into game-like activities can improve motivation and satisfaction of the learning session. This approach not only makes learning more enjoyable but also promotes active learning, thereby potentially improving educational outcomes. The project sets a precedent for future innovations in e-learning, illustrating the vast potential of integrating immersive and interactive elements in educational platforms to cater to the evolving needs of digital learners.
Adoption of technologies and practices from greek e-shops.
(ΕΛ.ΜΕ.ΠΑ., ΔΠΜΣ Οργάνωση και Διοίκηση για Μηχανικούς, 2021-11-19) Tzagaki, Eleni; Τζαγάκη, Ελένη
This thesis studies the level of adoption of factors and practices that are developed and used by Greek e-commerce websites. The study was conducted in order to evaluate factors that affect User's experience during the buying process, technological factors, tools and marketing practices that companies use to attract visitors and gain brand equity. The factors mentioned were collected by the literature and categorized into three sections, with the first one concerning the factors that affect UX and evaluated heuristically by criteria specifically formed for each factor. The second section contains technological tools and practices being used by the websites, and the third section records metrics about the website performance including SEO, speed of load, conversions, etc. After having evaluated all the factors, we examined the data by using statistical analysis tools and came up with the results that showed the usage level of each factor and the correlation between them. What we concluded is that, in general Greek on-line stores focus on providing the right guidance to the visitor to navigate their website while giving elaborate information concerning the company and the products used. Meanwhile the adoption level of technology is lower than expected as technological tools are not used by most websites. At the end we conclude that there are three approaches of marketing and visitor attracting practices that Greek e-commerce websites follow, with them being: price comparison websites presence, SCommerce adoption and classic advertising.
AI driven content recommendation systems.
(ΕΛΜΕΠΑ, Σχολή Επιστημών Διοίκησης και Οικονομίας (ΣΕΔΟ), ΠΜΣ Διοίκηση και Ψηφιακός Μετασχηματισμός, 2025-04-08) Μπλαζαντωνάκη, Αικατερίνη; Blazantonaki, Aikaterini; Μαστοράκης, Γεώργιος; Mastorakis, Georgios
Τα συστήματα συστάσεων αποτελούν έναν από τους πιο κρίσιμους τομείς της τεχνητής νοημοσύνης και της μηχανικής μάθησης, διαδραματίζοντας καθοριστικό ρόλο στη διαμόρφωση της εμπειρίας των χρηστών σε διάφορες ψηφιακές πλατφόρμες. Από το ηλεκτρονικό εμπόριο και τις υπηρεσίες streaming, έως τα κοινωνικά δίκτυα και τις διαδικτυακές μηχανές αναζήτησης, τα συστήματα αυτά αναλύουν δεδομένα χρηστών και παρέχουν εξατομικευμένες προτάσεις με στόχο την αύξηση της αλληλεπίδρασης και της ικανοποίησης του χρήστη. Η παρούσα εργασία εξετάζει τις βασικές αρχές λειτουργίας των συστημάτων συστάσεων, εστιάζοντας στις κύριες μεθόδους που χρησιμοποιούνται, όπως το συνεργατικό φιλτράρισμα, το φιλτράρισμα βάσει περιεχομένου και τα υβριδικά συστήματα που συνδυάζουν τις δύο προσεγγίσεις για βελτιστοποιημένα αποτελέσματα. Η μελέτη αναλύει τα πλεονεκτήματα των συστημάτων αυτών, όπως η βελτίωση της εμπειρίας του χρήστη, η αύξηση της δέσμευσης και η ενίσχυση των επιχειρηματικών στρατηγικών, καθώς και τις προκλήσεις που προκύπτουν. Ιδιαίτερη έμφαση δίνεται στην προστασία των προσωπικών δεδομένων, στη διαφάνεια των αλγορίθμων και στο φαινόμενο του «φίλτρου φούσκας», όπου οι χρήστες εκτίθενται μόνο σε ομοιογενές περιεχόμενο, περιορίζοντας την ποικιλία των πληροφοριών που λαμβάνουν. Επιπλέον, παρουσιάζονται μελέτες περιπτώσεων από δημοφιλείς πλατφόρμες όπως το Netflix, η Amazon, το YouTube και το Spotify, αναλύοντας τις στρατηγικές που εφαρμόζουν και τον τρόπο με τον οποίο χρησιμοποιούν τεχνικές βαθιάς μάθησης και νευρωνικών δικτύων για τη βελτίωση των συστάσεων. Τέλος, η εργασία εξετάζει τις μελλοντικές τάσεις στον τομέα των συστημάτων συστάσεων, με έμφαση σε καινοτομίες όπως το federated learning, το οποίο επιτρέπει την εξατομίκευση των προτάσεων χωρίς παραβίαση της ιδιωτικότητας των χρηστών, καθώς και νέες προσεγγίσεις για τη μείωση της μεροληψίας των αλγορίθμων.
Air quality evaluation using low-cost portable sensor node in indoor environment
(ΕΛΜΕΠΑ, Σχολή Μηχανικών (ΣΜΗΧ), ΠΜΣ Μηχανικών Πληροφορικής, 2024-04-18) Diamantakis, Ioannis; Διαμαντάκης, Ιωάννης; Panagiotakis, Spyridon; Παναγιωτάκης, Σπυρίδων
Air quality in large areas and especially in cities has been an issue of concern to the state and residents for decades, with the aim of ascertaining the situation and improving any problems that may arise, having developed common standards (AQI), surveillance and recording systems for decades. On the other hand, air quality in the places where we live, work, study, is not a key priority nor are there commonly accepted indicators by which the air of an enclosed space can be characterized as clean or harmful. Also, it is a fact that enclosed spaces differ significantly in air quality even, depending on their use even if they are very close. This creates the need for a large number of measuring stations and necessarily cheap ones to cover all spaces. In this thesis, a record is made of the methods of air measurement, the elements they take into account and a method of assessing indoor air quality (IAQ) is proposed, taking into account the burden caused by human presence. In addition, a device that uses the ESP32 microcontroller and a set of low cost, high accuracy sensors is designed, developed and tested. Its main purpose is to sense and send raw data of air metrics, using MQTT protocol, to a timeseries database for further processing. This device could be used to create a fog computing/sensor group, which could extend in a house or a larger building. It also proposes and develops a way of central monitoring (Grafana) of measurement stations and proposes a method of calculating the air quality index, with a centralized system and the use of machine learning algorithms like K-NN, that calculate the IAQ index of each station thus the room that is placed in, informing the inhabitants. The proposal is interesting and shows prospects for implementation. In the future, it could be expanded by adding automatic positioning of a measuring station, immediate suggestions for improving air quality depending on the pollutant that is increased, or controlling devices that could improve air quality.
Airbnb reviews sentiment and regression analysis.
(ΕΛΜΕΠΑ, Σχολή Μηχανικών (ΣΜΗΧ), ΠΜΣ Μηχανικών Πληροφορικής, 2025-01-21) Pirounakis, Georgios; Πηρουνάκης, Γεώργιος; Papadakis, Nikolaos; Παπαδάκης, Νικόλαος
The sharing economy represents a contemporary socio-economic model wherein individuals lease out their personal possessions, such as private cars or rooms in their homes, for short durations. This research endeavors to examine the factors influencing customer satisfaction in the utilization of proprietary rental websites within the sharing economy. Extensive datasets from Airbnb, comprising online reviews and listings in Crete, were subjected to analysis employing sentiment analysis, ordinal logistic regression, and visualization techniques. The results uncover that the sentiment of Airbnb guest reviews in Crete is significantly influenced by various factors, including hosts cleanliness, host communication, property type, host’s response rate and the host's superhost and identity verified status. Furthermore, positive expressions of satisfaction among Airbnb guests in Crete primarily revolved around the property's type and host interaction, followed by the amenities surrounding the property, cleanliness, and room quality. Conversely, negative online reviews were predominantly associated with also property's type/location followed by challenges related to insufficient host interaction, location, and room quality. The findings underscore the importance for Airbnb hosts to provide transparent and seamless information and room quality services, emphasizing effective communication channels with guests to enhance their experience and elevate satisfaction levels. Future studies should explore the generalizability of these findings to other urban contexts.
Aurora machinery: computer application for interactive audiovisual performance.
(ΕΛΜΕΠΑ, Σχολή Μουσικής και Οπτοακουστικών Τεχνολογιών (ΣΜΟΤ), ΠΜΣ Τεχνολογίες Ήχου και Μουσικής, 2024-09-03) Georgiopoulos, Anastasios; Γεωργιόπουλος, Αναστάσιος; Valsamakis, Nikolaos; Βαλσαμάκης, Νικόλαος
‘Aurora Machinery’ is system designed in the visual coding environment of Max/MSP to implement an audiovisual interactive installation, where the audience can interact and affect the sound output with the motion of their bodies. This thesis focuses not just on the examination of such a system and its creative use in composition. It provides a relevant chronological retrospection from anthropological, aesthetic and technological standpoint, delving into theoretical primary concerns such as how humans relate to music, the interconnection between music and society, the potential of technology as a creative tool, and the blur limits at which it becomes an end in itself.
Automated network data-driven seed generation for gray-box fuzzers based on generative adversarial network.
(ΕΛΜΕΠΑ, Σχολή Μηχανικών (ΣΜΗΧ), ΠΜΣ Μηχανικών Πληροφορικής, 2024-09-03) Kefaloukos, Ioannis-Georgios; Κεφαλούκος, Ιωάννης-Γεώργιος; Markakis, Evangelos; Μαρκάκης, Ευάγγελος
Cyber-attacks are assaults launched by cybercriminals utilizing network-enabled entities. Nowadays, the ever-increasing number of heterogeneous internet-connected entities has resulted in a wider cyber-threat landscape due to the exposure of their resources, services, and interfaces, potentially rendering the network prone to malicious attacks. With the introduction of new and the updation of existing architectural paradigms and methodologies such as Cloud Computing (CC), Fog Computing (FC), Extreme Edge Computing (EC), Internet of Things (loT), and Machine Learning (ML) and with the abundance of Information and Communications Technology (ICT) resources the attackers have devised more sophisticated ways to achieve their objectives. In our current era, the complexity that occurs in an ICT environment leads to new vulnerabilities within a network. At the same time, modern cybersecurity tools such as Intrusion Detection Systems (IDS), Anomaly Detection (AD), etc. focus on generic vulnerability categories. These issues, combined with the increase in variety and volume of cyber-attacks, pose an immense threat to everyone. Securing every connected entity within a network is a time-consuming, tedious, and challenging task, thus, a way to detect and address new vulnerabilities and bugs that each device and service contains had to be devised. Evidently, the detection of these vulnerabilities and bugs has to be automated, considering the number of services that co-exist in an entity and the vast number of different entities that co-exist within a network, thus guaranteeing its smooth operation. For that reason. fuzzing, an automated software testing technique, has gained traction for its potential to identify known and unknown vulnerabilities. However, prevailing fuzzing approaches often require human intervention in various steps of the process such as seed file generation, provide limited code coverage, and are resource-intensive. This work aims to explore automating and fine-tuning the seed file generation within grey-box fuzzers. To overcome these challenges, we propose a solution that automates seed refinement utilising a Generative Adversarial Network (GAN) in conjunction with dynamic network monitoring. By producing refined seed files and prioritizing them for testing, our approach aims to enhance vulnerability detection capabilities, deepening coverage while reducing processing time.
ByzElxeis: Ανάπτυξη ηλεκτρονικού εκπαιδευτικού εργαλείου, για υπολογισμό και ηχητική αναπαραγωγή των φθόγγων των κλιμάκων της Βυζαντινής μουσικής και της συμπεριφοράς τους κατά την εμφάνιση μελωδικών έλξεων, σε περιβάλλον Max/MSP.
(ΕΛ.ΜΕ.ΠΑ., Σχολή Μουσικής και Οπτοακουστικών Τεχνολογιών (ΣΜΟΤ), ΠΜΣ Τεχνολογίες Ήχου και Μουσικής, 2022-07-06) Συσκάκης, Αριστείδης; Syskakis, Aristeidis
Η εφαρμογή ByzElxeis είναι ένα μουσικό πρόγραμμα σε περιβάλλον Max/MSP, που αναπτύχθηκε ως ένα εκπαιδευτικό εργαλείο με σκοπό την υποστήριξη της μαθησιακής διαδικασίας κατά την σπουδή της βυζαντινής μουσικής. Εστιάζει στους φθόγγους και τα διαστήματα των θεωρητικών κλιμάκων των διατονικών Ήχων και ιδιαίτερα στις μικροτονικές μεταβολές ορισμένων φθόγγων κατά την εμφάνιση του φαινομένου των μελωδικών έλξεων. Η εφαρμογή υπολογίζει μαθηματικά τις ακριβείς συχνότητες των ηχομορίων που χρησιμοποιεί η βυζαντινή μουσική σε συγκερασμό 72 τμημάτων από επιλεγμένη τονική βάση. Χρησιμοποιώντας τις υπολογισμένες συχνότητες αναπαράγει τις κλίμακες των διατονικών Ήχων σε ανιούσα – κατιούσα μορφή. Για την παρουσίαση των μεταβολών συγκεκριμένων φθόγγων κατά την εμφάνιση μελωδικών έλξεων αναπαράγει χαρακτηριστικές μελωδικές φράσεις.
Charge energy dynamics investigation by Transient Absorption Spectroscopy on polymer composites with Single Wall Carbon Nanotubes and Ionic Liquid fillers, for thermoelectric applications.
(ΕΛΜΕΠΑ, Σχολή Μηχανικών (ΣΜΗΧ), ΔΠΜΣ Νανοτεχνολογία για Ενεργειακές Εφαρμογές, 2024-10-31) Rasouli, Angeliki; Ρασούλη, Αγγελική; Stratakis, Emmanouil; Konidakis, Ioannis; Στρατάκης, Εμμανουήλ; Κονιδάκης, Ιωάννης
In light of the ongoing demand for more efficient energy consumption and more sustainable energy sources, thermoelectric (TE)-based energy conversion is becoming increasingly significant. Polymer-based composites have grown in popularity in recent years, since they provide far more ecologically friendly options, they are far less expensive, easier to produce and process, and have innately low thermal conductivity. In the present work, polymer composites based on polycarbonate (PC) filled with, single-walled carbon nanotubes (SWCNTs, 0.5 wt %) and ionic liquids (ILs, 0.5 , 3.0 wt %) were studied under the implementation of transient absorption spectroscopy (TAS), in order to shed light on linking the signal of Seebeck coefficient (S) with the optical properties of the composites, under the influence of variant pump fluences. Also, scanning electron microscopy (SEM) was used, in order to study the morphology of the samples as far as their uniformity of their surface and CNT dispersion.
Chatbots και υποστήριξη πελατών στο ηλεκτρονικό εμπόριο.
(ΕΛΜΕΠΑ, Σχολή Επιστημών Διοίκησης και Οικονομίας (ΣΕΔΟ), ΠΜΣ Διοίκηση και Ψηφιακός Μετασχηματισμός, 2024-06-07) Μαυροειδή, Ουρανία; Mavroeidi, Ourania; Μαστοράκης, Γεώργιος; Mastorakis, Georgios
Η παρούσα ερευνητική μελέτη εξετάζει βασικές πτυχές των αλληλεπιδράσεων των χρηστών με chatbot σε περιβάλλοντα ηλεκτρονικού εμπορίου που αφορούν την εξατομίκευση, την οικοδόμηση εμπιστοσύνης και την επίδραση των προηγούμενων εμπειριών στις εμπειρίες των χρηστών chatbot, οι οποίες τελικά επηρεάζουν το αν θα χρησιμοποιήσουν chatbots στο μέλλον. Χρησιμοποιώντας μια ποσοτική μεθοδολογία, αναλύθηκαν δεδομένα από 122 χρήστες ηλεκτρονικού εμπορίου για να διερευνηθεί πώς αυτοί οι παράγοντες συμβάλλουν στην ικανοποίηση και την αφοσίωση των χρηστών απέναντι στα chatbots. Τα ευρήματα αποκαλύπτουν μια διαφοροποιημένη προοπτική όπου οι νεότεροι χρήστες παρουσιάζουν μεγαλύτερη ικανοποίηση από τις αλληλεπιδράσεις με chatbot σε σύγκριση με τις μεγαλύτερες ηλικίες. Η θετική προηγούμενη εμπειρία ενισχύει επίσης σημαντικά την προθυμία χρήσης του chatbot. Το επίπεδο εξατομίκευσης και η εμπιστοσύνη των χρηστών στα chatbots είχαν σημαντική, ισχυρή θετική σχέση με την προθυμία των χρηστών να χρησιμοποιήσουν το chatbot για να ολοκληρώσουν μια μελλοντική συναλλαγή. Αυτές οι γνώσεις σηματοδοτούν την ανάγκη για προσωπικές εμπειρίες chatbot, ώστε να υπάρχει εμπιστοσύνη προς την πλατφόρμα ηλεκτρονικού εμπορίου. Παρά τις γνώσεις που παρέχονται, η μελέτη αναγνωρίζει περιορισμούς, όπως η εξάρτηση από αυτοαναφερόμενα δεδομένα και η δειγματοληψία ευκολίας. Οι μελλοντικές έρευνες θα πρέπει να διεξάγουν διαχρονικές μελέτες που χρησιμοποιούν ποιοτικές μεθόδους για να εμβαθύνουν περισσότερο στις αποχρώσεις της αλληλεπίδρασης με τα chatbots. Συνεπώς, η παρούσα εργασία συμβάλλει στη βιβλιογραφία σχετικά με την ενσωμάτωση της ΤΝ στο ηλεκτρονικό εμπόριο με έμφαση στα σημεία κλειδιά που συμβάλλουν στον καθορισμό της εμπειρίας και της εμπλοκής των χρηστών με τα chatbots.
A cloud based key rolling technique for mitigating the man-in-the-middle attacks in LoRaWAN infrastructures
(ΕΛΜΕΠΑ, Σχολή Μηχανικών (ΣΜΗΧ), ΠΜΣ Μηχανικών Πληροφορικής, 2025-02-21) Astyrakakis, Nikolaos; Αστυρακάκης, Νικόλαος; Pallis, Evangelos; Πάλλης, Ευάγγελος
Recently, the Internet of Things (IoT) has emerged to a trending technology. IoT devices may capture, analyse, and process data before transferring it to application servers or sharing it through wired or wireless networks, across great distances. The LoRaWAN network is well-suited for such situations because it employs Low Power (LP) and Long Range (LoRa) protocols that reduce device energy consumption while maximising communication range. Before data transmission, all IoT devices on a LoRaWAN network must authenticate through a gateway to the backhaul, following an initial Join Procedure based on the submission of a Join Request. A Join Request includes two identifiers (AppEUI, DevEUI), a random value (DevNonce), a Media Access Control (MHDR) header, and a Message Integrity Code (MIC). The MIC value is the encryption result of either one of the aforementioned values or all of them together with an AppKey that is also stored in the backhaul for device authentication upon Join Request retrieval. Malicious actors acting as Man-In-the-Middle (MITM) can interfere in the channel and reverse engineer the AppKey; thereby, they may initiate a Join Request, which is misinterpreted as coming from a legitimate device, and eventually join the communication channel. This thesis presents a prototype technique that allows IoT devices to re-authenticate before every message exchange session, by constantly regenerating the AppKey. The suggested method uses a similar key rolling technique as the one used in automobile central locking systems, enhances the network's resistance against MITM attacks, and is developed as an optimized and scalable microservice for various LoRaWAN installations and versions.
Coverage-Based Summaries for RDF KBs.
(ΕΛ.ΜΕ.ΠΑ., Σχολή Μηχανικών (ΣΜΗΧ), ΠΜΣ Πληροφορική και Πολυμέσα, 2021-04-01) Vasileiou, Ioannis; Βασιλείου, Ιωάννης
As more and more data become available as linked data, the need for efficient and effective methods for their exploration becomes apparent. Semantic summaries try to extract meaning from data, while reducing its size. State of the art structural semantic summaries, focus primarily on the graph structure of the data, trying to maximize the summary’s utility for query answering, i.e. the query coverage. In this thesis, we present four algorithms, trying to maximize the aforementioned query coverage using ideas borrowed from result diversification. The key idea among all algorithms is, instead of focusing only to the “central” nodes, to push node selection also to the perimeter of the graph. Our experiments show the potential of our algorithms and demonstrate the considerable advantages gained for answering larger fragments of user queries.
Covid-19 και αυτοάνοσα νοσήματα.
(ΕΛ.ΜΕ.ΠΑ., Σχολή Επιστημών Υγείας (ΣΕΥ), ΠΜΣ Προηγμένη Κλινική Πρακτική στις Επιστήμες Υγείας, 2021-11-18) Χαριτάκη, Αριάδνη; Charitaki, Ariadni
Εισαγωγή. Στην παρούσα εργασία παρουσιάζουμε μια ανασκόπηση με στόχο την συσχέτιση μεταξύ COVID-19 και αυτοάνοσων ασθενειών, εστιάζοντας στις ομοιότητες ,στις κοινές ανοσολογικές αντιδράσεις, στην ανάπτυξη αυτοάνοσων ασθενειών σε ασθενείς με COVID-19 αλλά και στον κίνδυνο μόλυνσης από COVID19 σε ασθενείς με προϋπάρχουσες αυτοάνοσες καταστάσεις. Είναι προφανές, οτι κάποιοι άνθρωποι είναι πιο επιρρεπείς στη νόσο σε σχέση με κάποιους άλλους .Ο εμβολιασμός των ατόμων με αυτοάνοσα νοσήματα καθώς και οι παράγοντες που επιδρούν στην εμφάνιση σοβαρών συμπτωμάτων σε ασθενείς που νόσησαν από τον ιό SARS-CοV 2 θα μας απασχολήσουν. Σκοπός. Σκοπός της παρούσας μεταπτυχιακής εργασίας είναι η διερεύνηση της συσχέτισης του ιού covid -19 με τα αυτοάνοσα νοσήματα. Μεθοδολογία. Η προτεινόμενη ανασκόπηση αποτελεί μια περιγραφική ανασκόπηση. Οι βάσεις που χρησιμοποιήθηκαν για την αναζήτηση δεδομένων είναι το PUBMED και το GOOGLE SCHOLAR με βάση τον αλγόριθμο Boolean (AND/OR/NOT). Τα φίλτρα που χρησιμοποιήθηκαν για την αναζήτηση δεδομένων είναι η χρονική περίοδος δημοσίευσης των άρθρων(2015-2021),η γλώσσα. Τα άρθρα είναι δημοσιευμένα σε ελληνική και αγγλική γλώσσα. Η αναζήτηση ακολούθησε συγκεκριμένα κριτήρια και διεξήχθη με τη χρήση συγκεκριμένων όρων MESH. Αποτελέσματα. Η ανασκόπηση έδειξε ότι η ηλικία το αρσενικό φύλο και η συννοσηρότητα σχετίζονται με σοβαρές μορφές COVID-19.Δεν υπήρχαν σημαντικές διαφορές στη θνησιμότητα μεταξύ των ασθενών που νόσησαν από τον ιό SARS-CoV 2 οι οποίοι είχαν διαγνωστεί με αυτοάνοση νόσο σε σχέση με εκείνους που δεν είχαν αυτοάνοση νόσο. Oι ασθενείς που λάμβαναν κορτικοειδή ≥10 mg ανά ημέρα συσχετίστηκαν με αυξημένη πιθανότητα να νοσήσουν από τον ιό SARS -CoV-2. Οι ασθενείς με αυτοάνοσες ρευματικές παθήσεις μπορούν να εμβολιαστούν κατά του COVID-19. Ωστόσο συνίσταται ο εμβολιασμός σε αυτά τα άτομα όταν η δραστηριότητα της νόσου είναι υπό έλεγχο και όταν δεν υπάρχει ταυτόχρονη λοίμωξη. Ορισμένες θεραπείες αυτοάνοσων νοσημάτων μπορούν να μειώσουν την ανοσοαπόκριση των εμβολίων.H αποτελεσματικότητα των εμβολίων αποδείχθηκε χαμηλότερη στα άτομα με πολλαπλές συννοσηρότητες και στους ανοσοκατεσταλμένους. Συνεπώς συνιστάται διακοπή ορισμένων φαρμάκων πριν τον εμβολιασμό και επανέναρξη αυτών κάποιες εβδομάδες μετά. Τέλος αντιπυρηνικά αντισώματα (ΑΝΑs) ανιχνεύονται τόσο στους ασθενείς με αυτοάνοσα νοσήματα όσο και στους ασθενείς με COVID -19 . Τα αντιπυρηνικά αντισώματα (ANA) και τα αντιφωσφολιπιδικά αντισώματα (APL) είναι παραδείγματα αυτοαντισωμάτων που ανιχνεύονται σε αυτοάνοσες ασθένειες αλλά και σε ασθενείς με COVID19.Αυτό υποδηλώνει ότι ο ιός SARS-CoV-2 μπορεί να προκαλέσει αυτοανοσία. Συμπεράσματα. Ο COVID -19 μαζί με τα αυτοάνοσα νοσήματα αποτελούν συστηματικές νόσους καθώς μπορούν να επηρεάσουν και να προσβάλουν πολλά συστήματα ή όργανα του ανθρώπινου σώματος. Η διερεύνηση της σχέσης μεταξύ αυτών των ασθενειών είναι ζωτικής σημασίας για την πρόληψη και τη θεραπεία του COVID-19, καθώς και για την κατανόηση των κινδύνων για άτομα που ζουν με αυτοάνοσα νοσήματα.
Cyber security analysis on a production environment featuring software defined networks.
(ΕΛ.ΜΕ.ΠΑ., Σχολή Μηχανικών (ΣΜΗΧ), ΠΜΣ Πληροφορική και Πολυμέσα, 2020-10-09) Aspradakis, Panagiotis; Ασπραδάκης, Παναγιώτης
The ever-changing cyber threats are a defining feature of today's global landscape. Cyber Security is essential for organizations, businesses and many other stakeholders and one of the major issues for Industry 4.0. With a variety of people participating in organizations, businesses networks, industries, browsing the web, going about their business tasks or other online activities, steps must be taken to ensure the supervision of each user. The problem is that with the conventional network currently in use, a network administrator or an IT practitioner cannot have an accurate real-time view of how many devices are in the organization, business or industry, and does not have the flexibility to change the network topology on the fly, to address the real-time usage requirements or security incidents. SDN is very important for network organisation in Industry 4 because it enables separation of control from data flows. Hence it is inline with the other principles of softwarization and cloudification of the production. The controller orchestrates the network sessions between containerised services and end users or end devices. A firewall at the entrance of an enterprise network monitors the data traffic and enables rules enforcement at flow level. IDS/IPS enables the application of rules concerning protection of the infrastructure from malicious intruders and orchestrates the cybersecurity protection of the overall ecosystem from attackers. Motivated by the above, this Master’s Thesis will leverage the flexibility of Software Defined Networks (SDN), Deep Packet Inspection (DPI) technology and Network Function Virtualization (NFV), exploiting virtual machines to emulate a real production environment consisting of various production systems and security functions. Focusing on cybersecurity aspects, both the expanded attack surface that these new technologies introduced will be assessed, as well as the enhanced security mechanisms that they enable (e.g. reactive or moving defense techniques). To learn more about how SDN and related technologies work, and how someone with the administrative position over it can make it safer, we came up with the idea of setting up a virtual SDN network and implementing security solutions on it, seeing how it would behave while noting traffic and packet flow statistics. Since the virtualization platform that we used does not come with any SDN specific security tools or applications, we incorporated a traditional security feature into my implementation of the infrastructure. Legacy networks are also obsolete, providing a wide array of legacy security resources that one might use. SDN operates very similarly to a standard network, and when it comes to packet traffic and protocols, use of these resources on an SDN infrastructure is and is possible. The entire project then started to address around this, checking the efficient incorporation of existing network security tools into our functions to make SDN safer and more attack-proof.
Data warehousing in higher education. A case study of the Hellenic Mediterranean University.
(ΕΛ.ΜΕ.ΠΑ., ΔΠΜΣ Οργάνωση και Διοίκηση για Μηχανικούς, 2020-06-24) Smyrnaki, Ourania; Σμυρνάκη, Ουρανία
Business Information systems, such as ERP, CRM, MIS, etc., serve different needs in management, in customer relations, in the supply chain, in production monitoring etc. Data extracted from heterogeneous data sources can be exploited by Business Intelligence (BI) systems to support useful decision-making processes in business (industry). Advanced systems, such as ETL (Extract, Transform, Load) tools, are used for the collection of this data in Data Warehouses (DWs), thereby facilitating the processes of decision-making in Decision Support Systems (DSS). The information systems of the Hellenic Mediterranean University support the daily operations of the Institution such as management, financial, education etc. and collect heterogeneous data that are often used by members of the academic community and for various purposes. However, for effective decision making, the integration of this data that exist in various data sources, is required. The Institution will benefit by drawing useful conclusions that will help in making the right decisions. This will become a useful information service for the Quality Assurance Unit and the educational leadership of the Institute.
Deep learning models applied to SC-OCTA data for chads VASC score prediction
(ΕΛΜΕΠΑ, Σχολή Μηχανικών (ΣΜΗΧ), ΠΜΣ Μηχανικών Πληροφορικής, 2024-06-21) Xirogianni, Marielena; Ξηρογιάννη, Μαριέλενα; Papadourakis, Georgios; Παπαδουράκης, Γεώργιος
Thesis is research to find a solution how can extract information from OCT-A medical images using deep learning and machine models. OCT -A describe medical information, anatomical features of retina area. OCT-A two-dimensional images in grayscale to used medical information about the diabetes condition of patients’ eyes are AngioPlex OCT angiography from ZEISS medical devices OCT imaging of retinal microvasculature. AngioPlex OCT angiography is a product from a 3D image OCT volume and each one for four difference OCT-A images descried a level from the original OCT in 2D images. Intelligent artificial model with deep learning and machine learning model that recognition from medical images or recognized medical opinions is a step for that. Split the problem in three different parts. First part using medical images with pretrain models to recognize VASC Score as result the negative and positive diabetic condition of patients’ eyes as state of art EffienceNet model the most efficient pretrain model statically with less medical data. The second part is to explore and create medical data to train models for segmentation as result to extract information about Faz and verses. In the second part implement a research in publishes reach articles and medical official certified medical online libraries to collect information how can use them. The formula to extract the biomarkers is not open or clear how extracted is a part that the medical company Zessis have without publish. For our experiment have available a dataset from 400 patients with the information of biomarkers in a excel. The third part is to explore the dataset and organize it for machine learning experiments. Numerical data limitation requires needs statistical data of patients and biomarkers to investigate training experiments of parallel correlation of biomarkers with medical images with more than one parameter of positive or negative diabetes outcome.
Derivatives market before and after the financial crisis.
(ΕΛ.ΜΕ.ΠΑ., ΔΠΜΣ Οργάνωση και Διοίκηση για Μηχανικούς, 2020-10-19) Georgakis, Spyridon; Γεωργάκης, Σπυρίδων
Volatility in financial markets plays a very important role in making or wrecking the fortunes of investors. The study of volatility has attracted growing attention by researchers and policy makers since it is a measurement of risk, especially during extreme conditions such as financial crisis. This thesis through four models based on the opening, closing, high and low prices analyses and compares the volatility before and after the financial crisis of 2008. The global financial crisis initiated in the United States in early 2008 and then the financial meltdown had spread to the rest of the world. The study has been conducted on NASDAQ-100, FTSE-100 spot price indices, E-mini NASDAQ-100 and FTSE-100 Index Future futures price indices. Furthermore, this thesis examines the existence of the ‘monthly effect’ on returns and in volatility in U.S. and UK spot and futures markets between January 2006 and March 2019. The data have been partitioned into two sub-periods which allowed us to test the presence of monthly effect on returns or in volatility over the periods of pre-crisis (pre-2008) and post-crisis (post-2008). The sample employed in this dissertation comprise 3333 daily observations on NASDAQ-100, FTSE-100 spot price indices, E-mini NASDAQ-100 and FTSE-100 Index Future futures price indices. The set period of study was from 3 January 2006 to 4 March 2019. The results show that a simple measure of volatility (defined as the logarithmic difference between the high and low prices) overestimates the other three volatility estimators. The means of volatility estimators seem to have higher values during post-2008 period compared to pre-2008 period. Furthermore, the results from an OLS model show that there is no January effect in the UK and the US during the entire period and the two sub-periods. Regarding the impact of January effect in volatility of spot and futures indices, the hypothesis of January effect in volatility is accepted for FTSE-100 cash and stock index futures markets over the sub-period 2006-2007.
Design and development of a web-based data visualization software for political tendency identification of Twitter’s users using Python Dash Framework.
(ΕΛ.ΜΕ.ΠΑ., Σχολή Μηχανικών (ΣΜΗΧ), ΠΜΣ Πληροφορική και Πολυμέσα, 2020-06-11) Britzolakis, Alexandros; Μπριτζολάκης, Αλέξανδρος
The rapid evolution of computers as well as the emerge of the internet brought a new era on the field of communication systems. Many individuals can instantly communicate with each other (through instant messaging or through a video conference). This technological breakthrough set the stage for the emerge of the first internet communities. As a result, social media platforms emerged, were a set of free services is provided such as the interactive communication, multimedia content uploading etc. This new type of communication shapes the way by which an opinion can be expressed. Twitter is the most popular microblogging platform since its users can post a text of 280 maximum characters for a variety of subjects such as famous brands - products, celebrities, prominent events including political ones. As a result, Twitter is a tool that politicians tend to use frequently as it is a source for obtaining voters. This master thesis presents a web-based application that will use Twitter's API in order to obtain the most recent Tweets of the top three Greek political leaders and thus identify their additional popularity. To achieve this a set sample of recent posted tweets will be obtained (e.g. 200 tweets) from their additional Twitter accounts. These tweets will be processed in order to extract structured and unstructured data and present them in a form of graph series through a web-page. The structure of this web-based application consists of a frontend part created with HTML/CSS and a back-end mechanism which is developed using Python and Dash framework for the visualization process as well as Tweepy module for the application’s intercommunication with Twitter’s servers in order to obtain the data. More specifically the extracted information will be the number of likes, re-tweets and characters per posted tweet as well as the number of followers where they have. Furthermore, sentiment analysis of the tweet’s text is identified and visualized, using the Greek version of SpaCy module and labeled according to their corresponding expressed emotion. The extracted data will be used in order to display a set of charts that will present a comparison between these three political leaders. The research purpose of this dissertation is to present an engineering perspective on what data can mined from Twitter, how these data can be useful in order to estimate a political result as well as well as presenting the capabilities of Python Dash framework, Tweepy and SpaCy modules.
Design and implementation of a Hotel Reservation Back-end application with customized Web Service support.
(ΕΛ.ΜΕ.ΠΑ., Σχολή Μηχανικών (ΣΜΗΧ), ΠΜΣ Πληροφορική και Πολυμέσα, 2023-07-06) Kalkanis, Kyriakos; Καλκάνης, Κυριάκος
There is no doubt that technology is an integral part of our lives and directly affects our everyday life since it finds use in all fields of science and beyond. The internet and computers have aided people in many ways, including communication, education, entertainment, and other aspects of their lives. However, these are not the only ways technology can improve our lives. Websites and mobile applications are important technological creations that, when combined with electronic devices, are the means by which people carry out their daily tasks and manage their businesses. Nowadays more and more people who own businesses resort to setting up websites or mobile applications in order to manage them more productively and effectively. In this way, technology gives people the opportunity to save time and money as they can manage everything through a single website or application. To put it differently, they don't have to worry about all those complex processes that used to be done manually. In recent years, there has been a significant demand for the development of online applications in the hotel sector, which has benefited the economies of several countries as a result of tourism, and as a result, there is an increased demand. In this master thesis, a Property Management System (PMS) for hotel management was designed and implemented using Spring Boot for the back end and AngularJS for the front end. The purpose of this master thesis is the creation of a web application that will simplify and speed up hotel management. More specifically, connected users depending on their role in the hotel will have access to specific parts of the application. The most important features of this web application are the reservation system, contract creation for tour operators, pricing, the display of some statistics with the help of charts, as well as the development of a customized web service that enables real-time communication between this PMS and a booking application for the online room selling.