Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
4.5
2.7
Journals
Applied Sciences
Special Issues
Artificial Intelligence Applications and Innovation
share announcement

Artificial Intelligence Applications and Innovation

A special issue of Applied Sciences (ISSN 2076-3417). This special issue belongs to the section "Computing and Artificial Intelligence".

Deadline for manuscript submissions: closed (31 October 2021) | Viewed by 49184

Printed Edition Available!
A printed edition of this Special Issue is available here.

Special Issue Editors

Prof. Dr. João M. F. Rodrigues

E-Mail Website
Guest Editor
NOVA LINCS and Instituto Superior de Engenharia (ISE) , University of the Algarve, 8005-139 Faro, Portugal
Interests: computer vision; human–computer interaction; human–machine cooperation; artificial intelligence
Special Issues, Collections and Topics in MDPI journals
Prof. Dr. Pedro J. S. Cardoso

E-Mail Website
Guest Editor
NOVA LINCS and Instituto Superior de Engenharia (ISE), Universidade do Algarve , 8005-139 Faro, Portugal
Interests: operational research; machine learning; data science
Special Issues, Collections and Topics in MDPI journals
Dr. Marta Chinnici

E-Mail Website
Guest Editor
ICT Division-HPC Lab, Department of Energy Technologies and Renewable Energy Sources (TERIN), ENEA C.R. Casaccia, 00123 Roma, Italy
Interests: data science; artificial intelligence; machine learning; energy efficiency; digitalization; digital twin; data center; infrastructure; HPC
Special Issues, Collections and Topics in MDPI journals

Special Issue Information

Dear Colleagues,

The idea of an intelligent machine has fascinated humans for centuries. But what is intelligence? Some define it has the capacity for learning, reasoning, understanding or, from a different perspective, the aptitude of grasping truths, relationships, facts, or meanings. Any of these perspectives requires the capacity to acquire data from the surrounding environment and possibly acting over that environment. In short, the building of more or less autonomous agents, served with sensors and actuators, capable of learning and producing educated answers was long foreseen.

New trends comprise, among other aspects, pervasive robotization, ubiquitous online data access, empowered edge computing, smart spaces, and digital ethics. These trends build the research on “Artificial Intelligence Applications and Innovation”, impacting our day-to-day life, our cities, and even our free time. Nevertheless, artificial intelligence (AI) is still closely associated with some popular misconceptions that cause the general public to either have unrealistic fears about it or to have unrealistic expectations about how it will change our workplace and life in general. It is important to show that such fears are unfounded and that new trends, innovations, technologies, and smart systems will be able to improve the way we live, benefiting society without replacing humans in their core activities. 

This Special Issue will delve into mutually dependent subfields including, but not restricted to machine learning, computer vision, data analysis, data science, big data, Internet of Things, sentiment analysis, natural language processing, privacy and ethics, and robotics. Accepted papers will build a comprehensive collection of research and development trends on contemporary “Artificial Intelligence Applications and Innovation” that will serve as a convenient reference for AI experts as well as newly arrived practitioners, introducing them to the field’s trends.

Prof. Dr. João M. F. Rodrigues
Prof. Dr. Pedro J. S. Cardoso
Dr. Marta Chinnici
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. Applied Sciences 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 2400 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.

Keywords

  • accessibility and fighting info-exclusion
  • aging technologies
  • artificial intelligence
  • augmented and/or mixed reality applications
  • big data
  • computational science
  • computer vision
  • data science
  • decision support systems
  • deep learning
  • human–computer interaction
  • machine learning
  • navigation and optimization of routes
  • operational research
  • smart cities
  • smart tourism

Published Papers (15 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:

Editorial

Jump to: Research, Review

4 pages, 173 KiB  
Editorial
Artificial Intelligence Applications and Innovations: Day-to-Day Life Impact
by João M. F. Rodrigues, Pedro J. S. Cardoso and Marta Chinnici
Appl. Sci. 2023, 13(23), 12742; https://doi.org/10.3390/app132312742 - 28 Nov 2023
Cited by 1 | Viewed by 1001
Abstract
The idea of an intelligent machine has fascinated humans for centuries [...] Full article
(This article belongs to the Special Issue Artificial Intelligence Applications and Innovation)

Research

Jump to: Editorial, Review

15 pages, 25950 KiB  
Article
Development of a Stock Price Prediction Framework for Intelligent Media and Technical Analysis
by Sibusiso T. Mndawe, Babu Sena Paul and Wesley Doorsamy
Appl. Sci. 2022, 12(2), 719; https://doi.org/10.3390/app12020719 - 12 Jan 2022
Cited by 8 | Viewed by 4107
Abstract
Equity traders are always looking for tools that will help them maximise returns and minimise risk, be it fundamental or technical analysis techniques. This research integrates tools used by equity traders and uses them together with machine learning and deep learning techniques. The [...] Read more.
Equity traders are always looking for tools that will help them maximise returns and minimise risk, be it fundamental or technical analysis techniques. This research integrates tools used by equity traders and uses them together with machine learning and deep learning techniques. The presented work introduces a South African-based sentiment classifier to extract sentiment from new headlines and tweets. The experimental work uses four machine learning models for fundamental analysis and six long short-term memory model architectures, including a developed encoder-decoder long short-term memory model for technical analysis. Data used in the experiments is mined and collected from news sites, tweets from Twitter and Yahoo Finance. The results from 2 experiments show an accuracy of 96% in predicting one of the major telecommunication companies listed on the JSE closing price movement while using the linear discriminant analysis model and an RMSE of 0.023 in predicting a significant telecommunication company closing price using encoder-decoder long short-term memory. These findings reveal that the sentiment feature contains an essential fundamental value, and technical indicators also help move closer to predicting the closing price. Full article
(This article belongs to the Special Issue Artificial Intelligence Applications and Innovation)
►▼ Show Figures

Figure 1

16 pages, 1233 KiB  
Article
Assessing the Relevance of Opinions in Uncertainty and Info-Incompleteness Conditions
by Gerardo Iovane, Riccardo Emanuele Landi, Antonio Rapuano and Riccardo Amatore
Appl. Sci. 2022, 12(1), 194; https://doi.org/10.3390/app12010194 - 25 Dec 2021
Cited by 1 | Viewed by 2395
Abstract
Researchers are interested in defining decision support systems that can act in contexts characterized by uncertainty and info-incompleteness. The present study proposes a learning model for assessing the relevance of probability, plausibility, credibility, and possibility opinions in the conditions above. The solution consists [...] Read more.
Researchers are interested in defining decision support systems that can act in contexts characterized by uncertainty and info-incompleteness. The present study proposes a learning model for assessing the relevance of probability, plausibility, credibility, and possibility opinions in the conditions above. The solution consists of an Artificial Neural Network acquiring input features related to the considered set of opinions and other relevant attributes. The model provides the weights for minimizing the error between the expected outcome and the ground truth concerning a given phenomenon of interest. A custom loss function was defined to minimize the Mean Best Price Error (MBPE), while the evaluation of football players’ was chosen as a case study for testing the model. A custom dataset was constructed by scraping the Transfermarkt, Football Manager, and FIFA21 information sources and by computing a sentiment score through BERT, obtaining a total of 398 occurrences, of which 85% were employed for training the proposed model. The results show that the probability opinion represents the best choice in conditions of info-completeness, predicting the best price with 0.86 MBPE (0.61% of normalized error), while an arbitrary set composed of plausibility, credibility, and possibility opinions was considered for deciding successfully in info-incompleteness, achieving a confidence score of 2.47±0.188 MBPE (1.89±0.15% of normalized error). The proposed solution provided high performance in predicting the transfer cost of a football player in conditions of both info-completeness and info-incompleteness, revealing the significance of extending the feature space to opinions concerning the quantity to predict. Furthermore, the assumptions of the theoretical background were confirmed, as well as the observations found in the state of the art regarding football player evaluation. Full article
(This article belongs to the Special Issue Artificial Intelligence Applications and Innovation)
►▼ Show Figures

Figure 1

21 pages, 2545 KiB  
Article
Modular Dynamic Neural Network: A Continual Learning Architecture
by Daniel Turner, Pedro J. S. Cardoso and João M. F. Rodrigues
Appl. Sci. 2021, 11(24), 12078; https://doi.org/10.3390/app112412078 - 18 Dec 2021
Cited by 2 | Viewed by 2711
Abstract
Learning to recognize a new object after having learned to recognize other objects may be a simple task for a human, but not for machines. The present go-to approaches for teaching a machine to recognize a set of objects are based on the [...] Read more.
Learning to recognize a new object after having learned to recognize other objects may be a simple task for a human, but not for machines. The present go-to approaches for teaching a machine to recognize a set of objects are based on the use of deep neural networks (DNN). So, intuitively, the solution for teaching new objects on the fly to a machine should be DNN. The problem is that the trained DNN weights used to classify the initial set of objects are extremely fragile, meaning that any change to those weights can severely damage the capacity to perform the initial recognitions; this phenomenon is known as catastrophic forgetting (CF). This paper presents a new (DNN) continual learning (CL) architecture that can deal with CF, the modular dynamic neural network (MDNN). The presented architecture consists of two main components: (a) the ResNet50-based feature extraction component as the backbone; and (b) the modular dynamic classification component, which consists of multiple sub-networks and progressively builds itself up in a tree-like structure that rearranges itself as it learns over time in such a way that each sub-network can function independently. The main contribution of the paper is a new architecture that is strongly based on its modular dynamic training feature. This modular structure allows for new classes to be added while only altering specific sub-networks in such a way that previously known classes are not forgotten. Tests on the CORe50 dataset showed results above the state of the art for CL architectures. Full article
(This article belongs to the Special Issue Artificial Intelligence Applications and Innovation)
►▼ Show Figures

Figure 1

12 pages, 724 KiB  
Article
On the Design of a Decision Support System for Robotic Equipment Adoption in Construction Processes
by Carmen Marcher, Andrea Giusti and Dominik T. Matt
Appl. Sci. 2021, 11(23), 11415; https://doi.org/10.3390/app112311415 - 2 Dec 2021
Cited by 3 | Viewed by 1711
Abstract
The construction sector is one of the major global economies and is characterised by low productivity and high inefficiencies, but could highly benefit from the introduction of robotic equipment in terms of productivity, safety, and quality. As the development and the availability of [...] Read more.
The construction sector is one of the major global economies and is characterised by low productivity and high inefficiencies, but could highly benefit from the introduction of robotic equipment in terms of productivity, safety, and quality. As the development and the availability of robotic solutions for the construction sector increases, the evaluation of their potential benefits compared to conventional processes that are currently adopted on construction sites becomes compelling. To this end, we exploit Bayesian decision theory and apply an axiomatic design guideline for the development of a decision-theoretic expert system that: (i) evaluates the utility of available alternatives based on evidence; (ii) accounts for uncertainty; and (iii) exploits both expert knowledge and preferences of the users. The development process is illustrated by means of exemplary use case scenarios that compare manual and robotic processes. A use case scenario that compares manual and robotic marking and spraying is chosen for describing the development process in detail. Findings show how decision making in equipment selection can be supported by means of dedicated systems for decision support, developed in collaboration with domain experts. Full article
(This article belongs to the Special Issue Artificial Intelligence Applications and Innovation)
►▼ Show Figures

Graphical abstract

20 pages, 2247 KiB  
Article
Real Driving Cycle-Based State of Charge Prediction for EV Batteries Using Deep Learning Methods
by Seokjoon Hong, Hoyeon Hwang, Daniel Kim, Shengmin Cui and Inwhee Joe
Appl. Sci. 2021, 11(23), 11285; https://doi.org/10.3390/app112311285 - 29 Nov 2021
Cited by 10 | Viewed by 2897
Abstract
An accurate prediction of the State of Charge (SOC) of an Electric Vehicle (EV) battery is important when determining the driving range of an EV. However, the majority of the studies in this field have either been focused on the standard driving cycle [...] Read more.
An accurate prediction of the State of Charge (SOC) of an Electric Vehicle (EV) battery is important when determining the driving range of an EV. However, the majority of the studies in this field have either been focused on the standard driving cycle (SDC) or the internal parameters of the battery itself to predict the SOC results. Due to the significant difference between the real driving cycle (RDC) and SDC, a proper method of predicting the SOC results with RDCs is required. In this paper, RDCs and deep learning methods are used to accurately estimate the SOC of an EV battery. RDC data for an actual driving route have been directly collected by an On-Board Diagnostics (OBD)-II dongle connected to the author’s vehicle. The Global Positioning System (GPS) data of the traffic lights en route are used to segment each instance of the driving cycles where the Dynamic Time Warping (DTW) algorithm is adopted, to obtain the most similar patterns among the driving cycles. Finally, the acceleration values are predicted from deep learning models, and the SOC trajectory for the next trip will be obtained by a Functional Mock-Up Interface (FMI)-based EV simulation environment where the predicted accelerations are fed into the simulation model by each time step. As a result of the experiments, it was confirmed that the Temporal Attention Long–Short-Term Memory (TA-LSTM) model predicts the SOC more accurately than others. Full article
(This article belongs to the Special Issue Artificial Intelligence Applications and Innovation)
►▼ Show Figures

Figure 1

19 pages, 2820 KiB  
Article
Long Short-Term Memory Network-Based Metaheuristic for Effective Electric Energy Consumption Prediction
by Simran Kaur Hora, Rachana Poongodan, Rocío Pérez de Prado, Marcin Wozniak and Parameshachari Bidare Divakarachari
Appl. Sci. 2021, 11(23), 11263; https://doi.org/10.3390/app112311263 - 27 Nov 2021
Cited by 75 | Viewed by 2229
Abstract
The Electric Energy Consumption Prediction (EECP) is a complex and important process in an intelligent energy management system and its importance has been increasing rapidly due to technological developments and human population growth. A reliable and accurate model for EECP is considered a [...] Read more.
The Electric Energy Consumption Prediction (EECP) is a complex and important process in an intelligent energy management system and its importance has been increasing rapidly due to technological developments and human population growth. A reliable and accurate model for EECP is considered a key factor for an appropriate energy management policy. In recent periods, many artificial intelligence-based models have been developed to perform different simulation functions, engineering techniques, and optimal energy forecasting in order to predict future energy demands on the basis of historical data. In this article, a new metaheuristic based on a Long Short-Term Memory (LSTM) network model is proposed for an effective EECP. After collecting data sequences from the Individual Household Electric Power Consumption (IHEPC) dataset and Appliances Load Prediction (AEP) dataset, data refinement is accomplished using min-max and standard transformation methods. Then, the LSTM network with Butterfly Optimization Algorithm (BOA) is developed for EECP. In this article, the BOA is used to select optimal hyperparametric values which precisely describe the EEC patterns and discover the time series dynamics in the energy domain. This extensive experiment conducted on the IHEPC and AEP datasets shows that the proposed model obtains a minimum error rate relative to the existing models. Full article
(This article belongs to the Special Issue Artificial Intelligence Applications and Innovation)
►▼ Show Figures

Figure 1

24 pages, 6809 KiB  
Article
Deep Learning and Internet of Things for Beach Monitoring: An Experimental Study of Beach Attendance Prediction at Castelldefels Beach
by Mari Carmen Domingo
Appl. Sci. 2021, 11(22), 10735; https://doi.org/10.3390/app112210735 - 14 Nov 2021
Cited by 10 | Viewed by 2461
Abstract
Smart seaside cities can fully exploit the capabilities brought by Internet of Things (IoT) and artificial intelligence to improve the efficiency of city services in traditional smart city applications: smart home, smart healthcare, smart transportation, smart surveillance, smart environment, cyber security, etc. However, [...] Read more.
Smart seaside cities can fully exploit the capabilities brought by Internet of Things (IoT) and artificial intelligence to improve the efficiency of city services in traditional smart city applications: smart home, smart healthcare, smart transportation, smart surveillance, smart environment, cyber security, etc. However, smart coastal cities are characterized by their specific application domain, namely, beach monitoring. Beach attendance prediction is a beach monitoring application of particular importance for coastal managers to successfully plan beach services in terms of security, rescue, health and environmental assistance. In this paper, an experimental study that uses IoT data and deep learning to predict the number of beach visitors at Castelldefels beach (Barcelona, Spain) was developed. Images of Castelldefels beach were captured by a video monitoring system. An image recognition software was used to estimate beach attendance. A deep learning algorithm (deep neural network) to predict beach attendance was developed. The experimental results prove the feasibility of Deep Neural Networks (DNNs) for beach attendance prediction. For each beach, a classification of occupancy was estimated, depending on the number of beach visitors. The proposed model outperforms other machine learning models (decision tree, k-nearest neighbors, and random forest) and can successfully classify seven beach occupancy levels with the Mean Absolute Error (MAE), accuracy, precision, recall and F1-score of 0.03, 92.7%, 92.9%, 92.7%, and 92.7%, respectively. Full article
(This article belongs to the Special Issue Artificial Intelligence Applications and Innovation)
►▼ Show Figures

Figure 1

15 pages, 3344 KiB  
Article
Towards Facial Biometrics for ID Document Validation in Mobile Devices
by Iurii Medvedev, Farhad Shadmand, Leandro Cruz and Nuno Gonçalves
Appl. Sci. 2021, 11(13), 6134; https://doi.org/10.3390/app11136134 - 1 Jul 2021
Cited by 4 | Viewed by 2215
Abstract
Various modern security systems follow a tendency to simplify the usage of the existing biometric recognition solutions and embed them into ubiquitous portable devices. In this work, we continue the investigation and development of our method for securing identification documents. The original facial [...] Read more.
Various modern security systems follow a tendency to simplify the usage of the existing biometric recognition solutions and embed them into ubiquitous portable devices. In this work, we continue the investigation and development of our method for securing identification documents. The original facial biometric template, which is extracted from the trusted frontal face image, is stored on the identification document in a secured personalized machine-readable code. Such document is protected from face photo manipulation and may be validated with an offline mobile application. We apply automatic methods of compressing the developed face descriptors to make the biometric validation system more suitable for mobile applications. As an additional contribution, we introduce several print-capture datasets that may be used for training and evaluating similar systems for mobile identification and travel documents validation. Full article
(This article belongs to the Special Issue Artificial Intelligence Applications and Innovation)
►▼ Show Figures

Figure 1

27 pages, 4720 KiB  
Article
Investigating Issues and Needs of Dyslexic Students at University: Proof of Concept of an Artificial Intelligence and Virtual Reality-Based Supporting Platform and Preliminary Results
by Andrea Zingoni, Juri Taborri, Valentina Panetti, Simone Bonechi, Pilar Aparicio-Martínez, Sara Pinzi and Giuseppe Calabrò
Appl. Sci. 2021, 11(10), 4624; https://doi.org/10.3390/app11104624 - 19 May 2021
Cited by 16 | Viewed by 7233
Abstract
Specific learning disorders affect a significant portion of the population. A total of 80% of its instances are dyslexia, which causes significant difficulties in learning skills related to reading, memorizing and the exposition of concepts. Whereas great efforts have been made to diagnose [...] Read more.
Specific learning disorders affect a significant portion of the population. A total of 80% of its instances are dyslexia, which causes significant difficulties in learning skills related to reading, memorizing and the exposition of concepts. Whereas great efforts have been made to diagnose dyslexia and to mitigate its effects at primary and secondary school, little has been done at the university level. This has resulted in a sensibly high rate of abandonment or even of failures to enroll. The VRAIlexia project was created to face this problem by creating and popularizing an innovative method of teaching that is inclusive for dyslexic students. The core of the project is BESPECIAL, a software platform based on artificial intelligence and virtual reality that is capable of understanding the main issues experienced by dyslexic students and to provide them with ad hoc digital support methodologies in order to ease the difficulties they face in their academic studies. The aim of this paper is to present the conceptual design of BESPECIAL, highlighting the role of each module that composes it and the potential of the whole platform to fulfil the aims of VRAIlexia. Preliminary results obtained from a sample of about 700 dyslexic students are also reported, which clearly show the main issues and needs that dyslexic students experience and these will be used as guidelines for the final implementation of BESPECIAL. Full article
(This article belongs to the Special Issue Artificial Intelligence Applications and Innovation)
►▼ Show Figures

Figure 1

11 pages, 3000 KiB  
Article
Community Detection Based on Graph Representation Learning in Evolutionary Networks
by Dongming Chen, Mingshuo Nie, Jie Wang, Yun Kong, Dongqi Wang and Xinyu Huang
Appl. Sci. 2021, 11(10), 4497; https://doi.org/10.3390/app11104497 - 14 May 2021
Cited by 6 | Viewed by 1621
Abstract
Aiming at analyzing the temporal structures in evolutionary networks, we propose a community detection algorithm based on graph representation learning. The proposed algorithm employs a Laplacian matrix to obtain the node relationship information of the directly connected edges of the network structure at [...] Read more.
Aiming at analyzing the temporal structures in evolutionary networks, we propose a community detection algorithm based on graph representation learning. The proposed algorithm employs a Laplacian matrix to obtain the node relationship information of the directly connected edges of the network structure at the previous time slice, the deep sparse autoencoder learns to represent the network structure under the current time slice, and the K-means clustering algorithm is used to partition the low-dimensional feature matrix of the network structure under the current time slice into communities. Experiments on three real datasets show that the proposed algorithm outperformed the baselines regarding effectiveness and feasibility. Full article
(This article belongs to the Special Issue Artificial Intelligence Applications and Innovation)
►▼ Show Figures

Figure 1

18 pages, 3408 KiB  
Article
Convolutional Neural Networks for Differential Diagnosis of Raynaud’s Phenomenon Based on Hands Thermal Patterns
by Chiara Filippini, Daniela Cardone, David Perpetuini, Antonio Maria Chiarelli, Giulio Gualdi, Paolo Amerio and Arcangelo Merla
Appl. Sci. 2021, 11(8), 3614; https://doi.org/10.3390/app11083614 - 16 Apr 2021
Cited by 19 | Viewed by 2434
Abstract
Raynaud’s phenomenon (RP) is a microvessels’ disorder resulting in transient ischemia. It can be either primary or secondary to connective tissue diseases, such as systemic sclerosis. The differentiation between primary and secondary to systemic sclerosis is of paramount importance to set the proper [...] Read more.
Raynaud’s phenomenon (RP) is a microvessels’ disorder resulting in transient ischemia. It can be either primary or secondary to connective tissue diseases, such as systemic sclerosis. The differentiation between primary and secondary to systemic sclerosis is of paramount importance to set the proper therapeutic strategy. Thus far, thermal infrared imaging has been employed to accomplish this task by monitoring the finger temperature response to a controlled cold challenge. A completely automated methodology based on deep convolutional neural network is here introduced with the purpose of being able to differentiate systemic sclerosis from primary RP patients by relying uniquely on thermal images of the hands acquired at rest. The classification performance of such a method was compared to that of a three-dimensional convolutional neural network model implemented to classify thermal images of the hands recorded during rewarming from a cold challenge. No significant differences were found between the two procedures, thus ensuring the possibility to avoid the cold challenge. Moreover, the convolutional neural network models were compared with standard feature-based approaches and showed higher performances, thus overcoming the limitations related to the feature extraction (e.g., biases introduced by the operator). Such automated procedures can constitute promising tools for large scale screening of primary RP and secondary to systemic sclerosis in clinical practice. Full article
(This article belongs to the Special Issue Artificial Intelligence Applications and Innovation)
►▼ Show Figures

Figure 1

21 pages, 2899 KiB  
Article
Deep Data Assimilation: Integrating Deep Learning with Data Assimilation
by Rossella Arcucci, Jiangcheng Zhu, Shuang Hu and Yi-Ke Guo
Appl. Sci. 2021, 11(3), 1114; https://doi.org/10.3390/app11031114 - 26 Jan 2021
Cited by 52 | Viewed by 7598
Abstract
In this paper, we propose Deep Data Assimilation (DDA), an integration of Data Assimilation (DA) with Machine Learning (ML). DA is the Bayesian approximation of the true state of some physical system at a given time by combining time-distributed observations with a dynamic [...] Read more.
In this paper, we propose Deep Data Assimilation (DDA), an integration of Data Assimilation (DA) with Machine Learning (ML). DA is the Bayesian approximation of the true state of some physical system at a given time by combining time-distributed observations with a dynamic model in an optimal way. We use a ML model in order to learn the assimilation process. In particular, a recurrent neural network, trained with the state of the dynamical system and the results of the DA process, is applied for this purpose. At each iteration, we learn a function that accumulates the misfit between the results of the forecasting model and the results of the DA. Subsequently, we compose this function with the dynamic model. This resulting composition is a dynamic model that includes the features of the DA process and that can be used for future prediction without the necessity of the DA. In fact, we prove that the DDA approach implies a reduction of the model error, which decreases at each iteration; this is achieved thanks to the use of DA in the training process. DDA is very useful in that cases when observations are not available for some time steps and DA cannot be applied to reduce the model error. The effectiveness of this method is validated by examples and a sensitivity study. In this paper, the DDA technology is applied to two different applications: the Double integral mass dot system and the Lorenz system. However, the algorithm and numerical methods that are proposed in this work can be applied to other physics problems that involve other equations and/or state variables. Full article
(This article belongs to the Special Issue Artificial Intelligence Applications and Innovation)
►▼ Show Figures

Figure 1

13 pages, 841 KiB  
Article
Pre-Training Autoencoder for Lung Nodule Malignancy Assessment Using CT Images
by Francisco Silva, Tania Pereira, Julieta Frade, José Mendes, Claudia Freitas, Venceslau Hespanhol, José Luis Costa, António Cunha and Hélder P. Oliveira
Appl. Sci. 2020, 10(21), 7837; https://doi.org/10.3390/app10217837 - 5 Nov 2020
Cited by 12 | Viewed by 2947
Abstract
Lung cancer late diagnosis has a large impact on the mortality rate numbers, leading to a very low five-year survival rate of 5%. This issue emphasises the importance of developing systems to support a diagnostic at earlier stages. Clinicians use Computed Tomography (CT) [...] Read more.
Lung cancer late diagnosis has a large impact on the mortality rate numbers, leading to a very low five-year survival rate of 5%. This issue emphasises the importance of developing systems to support a diagnostic at earlier stages. Clinicians use Computed Tomography (CT) scans to assess the nodules and the likelihood of malignancy. Automatic solutions can help to make a faster and more accurate diagnosis, which is crucial for the early detection of lung cancer. Convolutional neural networks (CNN) based approaches have shown to provide a reliable feature extraction ability to detect the malignancy risk associated with pulmonary nodules. This type of approach requires a massive amount of data to model training, which usually represents a limitation in the biomedical field due to medical data privacy and security issues. Transfer learning (TL) methods have been widely explored in medical imaging applications, offering a solution to overcome problems related to the lack of training data publicly available. For the clinical annotations experts with a deep understanding of the complex physiological phenomena represented in the data are required, which represents a huge investment. In this direction, this work explored a TL method based on unsupervised learning achieved when training a Convolutional Autoencoder (CAE) using images in the same domain. For this, lung nodules from the Lung Image Database Consortium and Image Database Resource Initiative (LIDC-IDRI) were extracted and used to train a CAE. Then, the encoder part was transferred, and the malignancy risk was assessed in a binary classification—benign and malignant lung nodules, achieving an Area Under the Curve (AUC) value of 0.936. To evaluate the reliability of this TL approach, the same architecture was trained from scratch and achieved an AUC value of 0.928. The results reported in this comparison suggested that the feature learning achieved when reconstructing the input with an encoder-decoder based architecture can be considered an useful knowledge that might allow overcoming labelling constraints. Full article
(This article belongs to the Special Issue Artificial Intelligence Applications and Innovation)
►▼ Show Figures

Figure 1

Review

Jump to: Editorial, Research

35 pages, 2424 KiB  
Review
Re-Identification in Urban Scenarios: A Review of Tools and Methods
by Hugo S. Oliveira, José J. M. Machado and João Manuel R. S. Tavares
Appl. Sci. 2021, 11(22), 10809; https://doi.org/10.3390/app112210809 - 16 Nov 2021
Cited by 1 | Viewed by 2288
Abstract
With the widespread use of surveillance image cameras and enhanced awareness of public security, objects, and persons Re-Identification (ReID), the task of recognizing objects in non-overlapping camera networks has attracted particular attention in computer vision and pattern recognition communities. Given an image or [...] Read more.
With the widespread use of surveillance image cameras and enhanced awareness of public security, objects, and persons Re-Identification (ReID), the task of recognizing objects in non-overlapping camera networks has attracted particular attention in computer vision and pattern recognition communities. Given an image or video of an object-of-interest (query), object identification aims to identify the object from images or video feed taken from different cameras. After many years of great effort, object ReID remains a notably challenging task. The main reason is that an object’s appearance may dramatically change across camera views due to significant variations in illumination, poses or viewpoints, or even cluttered backgrounds. With the advent of Deep Neural Networks (DNN), there have been many proposals for different network architectures achieving high-performance levels. With the aim of identifying the most promising methods for ReID for future robust implementations, a review study is presented, mainly focusing on the person and multi-object ReID and auxiliary methods for image enhancement. Such methods are crucial for robust object ReID, while highlighting limitations of the identified methods. This is a very active field, evidenced by the dates of the publications found. However, most works use data from very different datasets and genres, which presents an obstacle to wide generalized DNN model training and usage. Although the model’s performance has achieved satisfactory results on particular datasets, a particular trend was observed in the use of 3D Convolutional Neural Networks (CNN), attention mechanisms to capture object-relevant features, and generative adversarial training to overcome data limitations. However, there is still room for improvement, namely in using images from urban scenarios among anonymized images to comply with public privacy legislation. The main challenges that remain in the ReID field, and prospects for future research directions towards ReID in dense urban scenarios, are also discussed. Full article
(This article belongs to the Special Issue Artificial Intelligence Applications and Innovation)
►▼ Show Figures

Figure 1

Show export options expand_more Show export options expand_less
Select all
Export citation of selected articles as:
 
Displaying articles 1-15
Back to TopTop