[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
You seem to have javascript disabled. Please note that many of the page functionalities won't work as expected without javascript enabled.
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
4.0
2.3
Journals
Mathematics
Special Issues
Ambient Intelligence Methods and Applications
share announcement

Ambient Intelligence Methods and Applications

A special issue of Mathematics (ISSN 2227-7390). This special issue belongs to the section "E1: Mathematics and Computer Science".

Deadline for manuscript submissions: closed (20 February 2025) | Viewed by 10673

Special Issue Editors

Dr. John Violos

E-Mail Website
Guest Editor
Department of Software and Information Technology Engineering, École de Technologie Supérieure, Montreal, QC, Canada
Interests: artificial intelligence; artificial neural network; data mining
Dr. Dimitrios Michail

E-Mail Website
Guest Editor
Department of Informatics and Telematics, Harokopio University of Athens, 17671 Athens, Greece
Interests: graph algorithms; graph mining; machine learning; algorithm engineering
Dr. Iraklis Varlamis

E-Mail Website
Guest Editor
Department of Informatics and Telematics, Harokopio University of Athens, Omirou 9, 17778 Athens, Greece
Interests: text mining; graph mining; social networks; healthcare; education
Special Issues, Collections and Topics in MDPI journals

Special Issue Information

Dear Colleague,

With data science and artificial intelligence becoming omnipresent and omnilayer, this has led researchers to reconsider and redesign our future cities, IoT devices and next-generation applications. This Special Issue has two purposes: First, it aims to present recent advances in data mining and machine learning techniques, focusing on knowledge discovery and optimal decision making with special emphasis on dynamic, heterogeneous and continuously changing environments. Second, it plans to present state-of-the-art data-driven applications that take input from distributed sources and build collective knowledge in order to improve the daily life and wellbeing of human society.

Sensors, the integration of actuators and processing units into our environment provides the contextual awareness and processing capability for revolutionizing how people interact and communicate with one another. This communication, also termed networking, includes humans, devices and appliances, giving novel opportunities to a wide range of telematic services and applications, such as autonomous vehicles, surveillance drones, energy-saving and e-health wearables, just to mention a few.

The evolution of human and animal species revealed that environmental factors have significantly influenced our intelligent lineage. Environmental changes and novel experiences have been found to be reflected in cognitive systems. In a similar way, we invite academic and industrial researchers to propose original, high-quality and innovative methods for incremental learning and few-shot and self-learning.

Topics of interest include, but are not limited to:

  1. Online learning;
  2. Zero- and few-shot learning;
  3. Self-learning systems;
  4. Distributed and collaborative learning;
  5. Data-driven IoT;
  6. Smart homes/cities/factoring/healthcare;
  7. Mobility and intelligent transportation;
  8. Intelligent networking;
  9. Drones for object and event detection;
  10. Cybernetics and human-to-machine interaction.

Dr. John Violos
Dr. Dimitrios Michail
Dr. Iraklis Varlamis
Guest Editors

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 100 words) can be sent to the Editorial Office for announcement on this website.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-blind peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Mathematics is an international peer-reviewed open access semimonthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 2600 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.

Benefits of Publishing in a Special Issue

  • Ease of navigation: Grouping papers by topic helps scholars navigate broad scope journals more efficiently.
  • Greater discoverability: Special Issues support the reach and impact of scientific research. Articles in Special Issues are more discoverable and cited more frequently.
  • Expansion of research network: Special Issues facilitate connections among authors, fostering scientific collaborations.
  • External promotion: Articles in Special Issues are often promoted through the journal's social media, increasing their visibility.
  • e-Book format: Special Issues with more than 10 articles can be published as dedicated e-books, ensuring wide and rapid dissemination.

Further information on MDPI's Special Issue policies can be found here.

Published Papers (3 papers)

Download All Papers
Order results
Result details
Show export options expand_more Show export options expand_less
Select all
Export citation of selected articles as:

Research

30 pages, 1422 KiB  
Article
A Comparative Analysis of Compression and Transfer Learning Techniques in DeepFake Detection Models
by Andreas Karathanasis, John Violos and Ioannis Kompatsiaris
Mathematics 2025, 13(5), 887; https://doi.org/10.3390/math13050887 - 6 Mar 2025
Viewed by 136
Abstract
DeepFake detection models play a crucial role in ambient intelligence and smart environments, where systems rely on authentic information for accurate decisions. These environments, integrating interconnected IoT devices and AI-driven systems, face significant threats from DeepFakes, potentially leading to compromised trust, erroneous decisions, [...] Read more.
DeepFake detection models play a crucial role in ambient intelligence and smart environments, where systems rely on authentic information for accurate decisions. These environments, integrating interconnected IoT devices and AI-driven systems, face significant threats from DeepFakes, potentially leading to compromised trust, erroneous decisions, and security breaches. To mitigate these risks, neural-network-based DeepFake detection models have been developed. However, their substantial computational requirements and long training times hinder deployment on resource-constrained edge devices. This paper investigates compression and transfer learning techniques to reduce the computational demands of training and deploying DeepFake detection models, while preserving performance. Pruning, knowledge distillation, quantization, and adapter modules are explored to enable efficient real-time DeepFake detection. An evaluation was conducted on four benchmark datasets: “SynthBuster”, “140k Real and Fake Faces”, “DeepFake and Real Images”, and “ForenSynths”. It compared compressed models with uncompressed baselines using widely recognized metrics such as accuracy, precision, recall, F1-score, model size, and training time. The results showed that a compressed model at 10% of the original size retained only 56% of the baseline accuracy, but fine-tuning in similar scenarios increased this to nearly 98%. In some cases, the accuracy even surpassed the original’s performance by up to 12%. These findings highlight the feasibility of deploying DeepFake detection models in edge computing scenarios. Full article
(This article belongs to the Special Issue Ambient Intelligence Methods and Applications)
►▼ Show Figures

Figure 1

Figure 1
<p>Pruning of convolutional neural networks.</p> Full article ">Figure 2
<p>Knowledge distillation in the teacher–student framework.</p> Full article ">Figure 3
<p>Quantization of deep neural network parameters.</p> Full article ">Figure 4
<p>Low-rank factorization.</p> Full article ">Figure 5
<p>Transfer learning across different tasks.</p> Full article ">Figure 6
<p>CNN with adapter module for transfer learning.</p> Full article ">Figure 7
<p>Knowledge distillation for transfer learning.</p> Full article ">Figure 8
<p>“Dogs vs. cats” dataset example.</p> Full article ">Figure 9
<p>Sample ROC curves for Synthbuster dataset.</p> Full article ">
24 pages, 5312 KiB  
Article
Transfer Learning for Day-Ahead Load Forecasting: A Case Study on European National Electricity Demand Time Series
by Alexandros Menelaos Tzortzis, Sotiris Pelekis, Evangelos Spiliotis, Evangelos Karakolis, Spiros Mouzakitis, John Psarras and Dimitris Askounis
Mathematics 2024, 12(1), 19; https://doi.org/10.3390/math12010019 - 21 Dec 2023
Cited by 4 | Viewed by 2153
Abstract
Short-term load forecasting (STLF) is crucial for the daily operation of power grids. However, the non-linearity, non-stationarity, and randomness characterizing electricity demand time series renders STLF a challenging task. Various forecasting approaches have been proposed for improving STLF, including neural network (NN) models [...] Read more.
Short-term load forecasting (STLF) is crucial for the daily operation of power grids. However, the non-linearity, non-stationarity, and randomness characterizing electricity demand time series renders STLF a challenging task. Various forecasting approaches have been proposed for improving STLF, including neural network (NN) models which are trained using data from multiple electricity demand series that may not necessarily include the target series. In the present study, we investigate the performance of a special case of STLF, namely transfer learning (TL), by considering a set of 27 time series that represent the national day-ahead electricity demand of indicative European countries. We employ a popular and easy-to-implement feed-forward NN model and perform a clustering analysis to identify similar patterns among the load series and enhance TL. In this context, two different TL approaches, with and without the clustering step, are compiled and compared against each other as well as a typical NN training setup. Our results demonstrate that TL can outperform the conventional approach, especially when clustering techniques are considered. Full article
(This article belongs to the Special Issue Ambient Intelligence Methods and Applications)
►▼ Show Figures

Figure 1

Figure 1
<p>Visual representation of the outlier removal and missing value imputation procedures that were performed during the pre-processing of the data set. (<b>a</b>) Extreme outliers at random data points (Denmark). (<b>b</b>) Outliers forming grouping at specific data points (Norway). (<b>c</b>) Imputation matching complicated pattern (France). (<b>d</b>) Imputation on multiple close proximity data points (Greece).</p> Full article ">Figure 2
<p>Average load profiles of the entire load time series data set aggregated by selected time intervals (hour, weekday, and month) for the time period of 2015 to 2021. (<b>a</b>) Average daily load profile (hourly resolution). (<b>b</b>) Average weekly load profile (daily resolution). (<b>c</b>) Average yearly load profile (monthly resolution).</p> Full article ">Figure 3
<p>An illustration of the derived load profile vectors containing daily and yearly profiles for each country of the data set.</p> Full article ">Figure 4
<p>Dendrogram and related choropleth based on clustering performed on daily, weekly, and yearly profiles of the countries.</p> Full article ">Figure 5
<p>An illustration of the separation process of source and target domains for each experiment in the AbO setup. Each row of cells represents a separate execution of the entire ML pipeline. Source countries in each row are depicted in green, while the corresponding target domain country is depicted in blue. The same approach has been followed in the CbO setup as well, with the exception that this separation takes place internally within the countries of each cluster, excluding those that do not pertain to it.</p> Full article ">Figure 6
<p>Side-by-side barplots depicting the MAPE (%) for each TL setup and baseline for each country.</p> Full article ">Figure 7
<p>Side-by-side barplots depicting the average MAPE (%) for each TL setup and the baseline for each cluster.</p> Full article ">Figure A1
<p>Illustration of the pipeline’s execution flow for the machine learning life-cycle stages.</p> Full article ">
25 pages, 875 KiB  
Article
An Intelligent Recommendation System for Automating Academic Advising Based on Curriculum Analysis and Performance Modeling
by Shadi Atalla, Mohammad Daradkeh, Amjad Gawanmeh, Hatim Khalil, Wathiq Mansoor, Sami Miniaoui and Yassine Himeur
Mathematics 2023, 11(5), 1098; https://doi.org/10.3390/math11051098 - 22 Feb 2023
Cited by 27 | Viewed by 7238
Abstract
The explosive increase in educational data and information systems has led to new teaching practices, challenges, and learning processes. To effectively manage and analyze this information, it is crucial to adopt innovative methodologies and techniques. Recommender systems (RSs) offer a solution for advising [...] Read more.
The explosive increase in educational data and information systems has led to new teaching practices, challenges, and learning processes. To effectively manage and analyze this information, it is crucial to adopt innovative methodologies and techniques. Recommender systems (RSs) offer a solution for advising students and guiding their learning journeys by utilizing statistical methods such as machine learning (ML) and graph analysis to analyze program and student data. This paper introduces an RS for advisors and students that analyzes student records to develop personalized study plans over multiple semesters. The proposed system integrates ideas from graph theory, performance modeling, ML, explainable recommendations, and an intuitive user interface. The system implicitly implements many academic rules through network analysis. Accordingly, a systematic and comprehensive review of different students’ plans was possible using metrics developed in the mathematical graph theory. The proposed system systematically assesses and measures the relevance of a particular student’s study plan. Experiments on datasets collected at the University of Dubai show that the model presented in this study outperforms similar ML-based solutions in terms of different metrics. Typically, up to 86% accuracy and recall have been achieved. Additionally, the lowest mean square regression (MSR) rate of 0.14 has been attained compared to other state-of-the-art regressors. Full article
(This article belongs to the Special Issue Ambient Intelligence Methods and Applications)
►▼ Show Figures

Figure 1

Figure 1
<p>Recommender system architecture.</p> Full article ">Figure 2
<p>Hypothetical curricula and the generated course prerequisite network.</p> Full article ">Figure 3
<p>The system’s functional architecture.</p> Full article ">Figure 4
<p>Overview of the proposed Bayesian belief networks predictive model.</p> Full article ">Figure 5
<p>Betweenness centrality of IT and concentration courses in BSCIS-ISS at the University of Dubai. Numbers and node height indicate each course’s betweenness centrality (0 and 1.0). The hilighted nodes indicate the longest path for curriculum.</p> Full article ">Figure 6
<p>GPA values predicted by the random forest regressor vs. the actual data.</p> Full article ">
Show export options expand_more Show export options expand_less
Select all
Export citation of selected articles as:
 
Displaying articles 1-3
Back to TopTop