Search

Many oversampling methods applied to imbalanced data generate samples according to local density distribution of minority samples. However, samples generated by these methods can only present a non-deterministic relationship between the local and global distributions. A generative adversarial network (GAN) is a suitable tool to learn an unknown global probability distribution. In this paper, we propose an improved GAN (I-GAN) to oversample according to the global underlying structure of minority samples. The originality of I-GAN stems from the fact it provides additional density distribution information of minority samples for GAN and generated samples. By building on this idea, three detailed strategies are presented: input random vectors of the generator are sampled from a rough estimate of the distribution of minority samples to orientate fake samples more believable; a residual about minority samples is added into the discriminator to strengthen the constraint of loss function; generated samples are redistributed with a reshaper. These three strategies provide innovative methodologies at various stages of GANs for the oversampling task. Compared with 22 classical and popular imbalanced sampling methods under metrics of G m , F 1 , and A U C on 24 benchmark imbalanced datasets, it is shown that I-GAN is effective and robust. The I-GAN implementation line procedure has been uploaded to Github (https://github.com/flowerbloom000/I-GAN). [ABSTRACT FROM AUTHOR]

A deep learning model can perform efficient uncertainty quantification (UQ) for reservoir flow with uncertain model parameters, but usually requires large amounts of training data to ensure accuracy. However, the cost of obtaining large amounts of data is prohibitive, and the performance will deteriorate if sufficient training data is lacking. Alternatively, more interpretable neural networks with embedded physical laws have recently been used to solve partial differential equations as well as to solve UQ problems. This approach has received a lot of attention due to its low data volume requirements and its adherence to the laws of physics during the training process. In this paper, we propose a theory-guided framework based on a bilevel programming model with hard constraints to embed physical meaning in the model. Theory guided Lagrange programming neural network (TGLPNN) combines the method of Lagrange programming neural network approach where physical laws such as stochastic partial differential equations and boundary conditions are incorporated into the training process of a convolutional neural network. At the same time, the upper-level variables are iteratively optimized. The method based on Lagrange programming neural network inherently embeds physical laws in the network. Practical applications have shown that TGLPNN can provide higher prediction accuracy compared to state-of-the-art physics-driven methods and improved efficiency compared to numerical methods. • Improving embedding theory guidance using Lagrange Programming neural network. • Optimizing hyperparameters using Meta-Weight-Net. • Effective neural network approach to subsurface flow problems. [ABSTRACT FROM AUTHOR]

Intelligent applications are highly susceptible to uncertainty and inconsistency due to the intense and intricate interactions among their autonomous components (or agents), making their verification theoretically and practically challenging. This paper presents the design and implementation of a new open-source and scalable software tool for modeling and verifying intelligent applications with commitment and trust protocols under both uncertainty and inconsistency settings, using reduction-based multi-valued model checking techniques. The proposed tool is equipped with original and novel algorithms that transform our logics of multi-valued commitment (mv-CTLC) and multi-value trust (mv-TCTL) that we recently introduced to their classical two-valued commitment (CTLC) and trust (TCTL) logic versions as well as to Computational Tree Logic (CTL). Moreover, the tool transforms the mv-CTL to CTL, and it is applicable for the classical model checking by transforming the classical logics of trust and commitment to CTL. To demonstrate the practicality and applicability of the proposed tool in real settings, we present and report experimental results over two blockchain-based applications in the healthcare domain. Finally, we provide discussions and comparisons between the proposed approaches regarding scalability and efficiency. Moreover, we provide packages of more than 11 experiments, including the ones we conduct in this paper and enhanced experiments from previous works. Our findings ensure that the proposed approaches and the software tool that implements them are highly efficient and scalable, giving accurate results under varying conditions. • Design and implementation of an open-source scalable tool for systems model checking. • System modeling with multi-valued logic capturing uncertainty and inconsistency. • Practical demonstration using blockchain-based applications in healthcare domain. • Extensive experiments showing scalability, efficiency and reliability of the tool. [ABSTRACT FROM AUTHOR]

Increases in population and prosperity are linked to a worldwide rise in garbage. The "classification" and "recycling" of solid waste is a crucial tactic for dealing with the waste problem. This paper presents a new two-layer intelligent decision system for waste sorting based on fused features of Deep Learning (DL) models as well as a selection of an optimal deep Waste-Sorting Model (WSM) based on Multi-Criteria Decision Making (MCDM). A dataset comprising 1451 samples of images of waste, distributed across four classes – cardboard (403), glass (501), metal (410), and general trash (137), was used for sorting. This study proposes a Multi-Fused Decision Matrix (MFDM) based on identified fusion score level rules, evaluation criteria, and deep fused waste-sorting models. Five fusion rules used in the sorting process and the evaluation perspectives into the MFDM are sum, weighted sum, product, maximum, and minimum rules. Additionally, each of entropy and Visekriterijumska Optimizacija i Kompromisno Resenje in Serbian (VIKOR) methods was used for weighting selected criteria as well as ranking deep WSMs. The highest accuracy rate of 98% was scored by ResNet50-GoogleNet- Inception based on the minimum rule. However, under the same rule, an insufficient accuracy rate of sorting was presented by ResNet50-GoogleNet-Xception. Since Qi = 0 for Inception-Xception, the final output based on MCDM methods indicates that the fused Inception-Xception model outperforms the other fused deep WSMs, which achieved the lowest values of Qi. Thus, Inception-Xception was chosen as the best deep waste-sorting model based on images of waste, multiple evaluation criteria, and different fusion perspectives. The mean and standard deviation metrics were both used to validate the selection findings objectively. The suggested approach can aid urban decision-makers in prioritizing and choosing an Artificial Intelligence (AI)-optimized optimal sorting model. • Develop a 2-layer decision system for waste sorting using deep learning. • Construct MFDM using fusion scores, criteria, and deep waste sorting models. • Entropy and VIKOR methods rank criteria and deep learning models. [ABSTRACT FROM AUTHOR]

Trust and social commitments have been studied with different objectives for communication in Multi-Agent Systems (MASs) separately. The purpose of this paper is to present the first logical framework to explore the relationship between two key concepts in autonomous MASs, namely trust and commitments among agents, along with its model checking algorithms. An analysis is performed on this relationship, with a specific emphasis on the perspectives of semantics and model checking. The analysis is carried out within the framework of Trust Computation Tree Logic with Commitments (CTLC) logic, an extended logic built on top of the standard branching temporal Computation Tree Logic (CTL) by including modalities for reasoning about commitments and their fulfillment. Moreover, the study employs a trust operator to facilitate trust reasoning. We propose thus Trust Computation Tree Logic with Commitments (TCTLC), a new logic able to express properties about trust and commitments simultaneously. Such rich properties cannot be expressed in other languages. We introduce the semantics using the extended formalism of interpreted systems. We propose some postulates and proofs that show how we can reason about combined trust and commitments in the logic proposed. New Binary Decision Diagram (BDD)-based algorithms for our logic are presented and implemented as additional libraries of the Model Checker for Multi-Agent Systems (MCMAS), the default model checker of MASs. Furthermore, we prove that although TCTLC extends CTL, its model checking algorithm remains polynomial (P-complete) with respect to the size of the model and the length of the formula. Similarly, the space complexity for concurrent programs also remains unchanged, which is polynomial (PSPACE-complete) with respect to the size of the components within these programs. Finally, to evaluate the proposed technique, we report the experimental results using an industrial case study and compare the results with those obtained on relevant benchmarks. • We introduce a new logical language combining trust and social commitments. • We introduce a new model checking algorithm for the new trust and commitment logic. • We propose reasoning postulates and proofs to reason about trust and commitments. • We implement the new algorithm and report its practical efficiency on a case study. [ABSTRACT FROM AUTHOR]

Transformers are Deep Neural Networks (DNN) that utilize a self-attention mechanism to capture contextual relationships within sequential data. Unlike traditional neural networks and variants of Recurrent Neural Networks (RNNs), such as Long Short-Term Memory (LSTM), Transformer models excel at managing long dependencies among input sequence elements and facilitate parallel processing. Consequently, Transformer-based models have garnered significant attention from researchers in the field of artificial intelligence. This is due to their tremendous potential and impressive accomplishments, which extend beyond Natural Language Processing (NLP) tasks to encompass various domains, including Computer Vision (CV), audio and speech processing, healthcare, and the Internet of Things (IoT). Although several survey papers have been published, spotlighting the Transformer's contributions in specific fields, architectural disparities, or performance assessments, there remains a notable absence of a comprehensive survey paper that encompasses its major applications across diverse domains. Therefore, this paper addresses this gap by conducting an extensive survey of proposed Transformer models spanning from 2017 to 2022. Our survey encompasses the identification of the top five application domains for Transformer-based models, namely: NLP, CV, multi-modality, audio and speech processing, and signal processing. We analyze the influence of highly impactful Transformer-based models within these domains and subsequently categorize them according to their respective tasks, employing a novel taxonomy. Our primary objective is to illuminate the existing potential and future prospects of Transformers for researchers who are passionate about this area, thereby contributing to a more comprehensive understanding of this groundbreaking technology. • The paper presents a comprehensive survey on transformers for deep learning tasks. • The paper conducts a thorough analysis on highly effective models in five domains. • The paper classifies the models based on respective tasks using a proposed taxonomy. • The characteristics of the surveyed models are deeply explored and analyzed. • Future directions and challenges for transformer-based models are deciphered. [ABSTRACT FROM AUTHOR]

A ratio of CO and CO 2 (CO / CO 2 ) is a key indicator of sintering carbon consumption, which is difficult to be determined in real-time. Therefore, the establishment of its soft sensing model is of great practical significance. This paper proposes a novel CO / CO 2 soft sensing model with incomplete output data based on relevance vector machine with hybrid kernel via data augmentation. First, a least absolute shrinkage and selection operator is employed for determining key input variables of the model, and an automatic fuzzy clustering framework is used to automatically identify multiple operating modes. Then, a relevance vector machine with hybrid kernel method is presented to model each operating mode separately. Meanwhile, considering the problem of incomplete input and output data, data augmentation is applied in modeling to enhance the model performance. Finally, the soft sensing model of CO / CO 2 is formed. Experimental results and analyses using actual production data coming from the sintering production process demonstrate that the prediction performance and accuracy of the proposed model outperform some existing algorithms. [ABSTRACT FROM AUTHOR]

Intuitionistic fuzzy (IF) theory has become main approach to representing imprecision and vagueness. The IF divergence measure (IFDivM) based on Jensen–Shannon divergence is perhaps the most widely used measure to compare the similarity of multiple intuitionistic fuzzy sets (IFSs). In the present paper, this IFDivM is examined and applied to multiple examples. It is found that some extant IFDivMs hardly satisfy the axiomatic definition, and in a few cases even unable to show divergence of trivial IFSs. To address these inconsistencies, a new IFDivM based on Jensen–Shannon divergence is proposed, free from these problems. The effectiveness of the proposed IFDivM is tested on several critical cases, and precise analysis of its properties is performed. It is proved that the proposed IFDivM satisfies the axiomatic definition of IFDivMs. To illustrate the practical significance of the IFDivM, a novel intuitionistic fuzzy (IF) TODIM method, based on the proposed IFDivM, is developed, termed as GIF-TODIM method. Unlike the existing IF-TODIM methods, GIF-TODIM does not suffer from the revere ordering inconsistencies. The proposed GIF-TODIM method and the proposed IFDivM are applied to a real-world case study on supplier selection. A detailed comparative analysis is performed taking the TOPSIS method and other IFDivMs as baselines. The role of attitude on the final choice is analyzed in great detail. It is found that the proposed GIF-TODIM method is indeed useful, effective, and superior to the counterpart methods, when it comes to real-world situations. Concomitantly, in the present work, it is also revealed that the TOPSIS method based on the 2-D Hamming distance is a special form of the proposed GIF-TODIM method, when decision-makers have the same attitude towards losses and gains. Thus, an interesting relationship between TOPSIS and TODIM is identified under the intuitionistic fuzzy environment, which is bound to propel significant research in the area of decision making under uncertain conditions. As a whole, the article offers comprehensive analyses of IFDivMs and the TODIM method under the intuitionistic fuzzy environment. • Construct a new IF divergence measure (IFDivM) based on Jensen–Shannon divergence. • Develop a GIF-TODIM method based on the new IFDivM. • Apply the proposed GIF-TODIM method to a supplier selection problem. • Provide parameter analysis and comparative analysis with TOPSIS and other IFDivMs. • Show that TOPSIS based on 2-D Hamming distance is a special form of our GIF-TODIM. [ABSTRACT FROM AUTHOR]

Fifth-generation (5G) technology provides new possibilities for a variety of applications, but it also comes with challenges influenced by distinct aspects, such as the size of organizations that use such technology. Therefore, it is important to understand which architecture of 5G-radio access networks (RANs) is best for a given purpose; this requires an evaluation platform for assessment. This paper tackles this problem by presenting a novel multi-criteria decision-making (MCDM) solution based on a new integrated fuzzy set. The proposed integrated approach, which is based on a Type-2 neutrosophic fuzzy environment, is developed to address the application challenges of 5G-RANs architecture evaluation, as also to face the MCDM theoretical challenge represented by ambiguities and inconsistencies among decision makers within the decision making context of the presented case study. Many MCDM techniques for weighting and selection were presented from the literature, yet many of them still suffer from inconsistencies and uncertainty. Therefore, the chosen methods in this research are unique in a way that previous issues are addressed, making them suitable for integration with Type-2 neutrosophic fuzzy environment, and therefore creating a more robust decision platform for the presented challenge in this research, as a theoretical contribution. First, a new "Type-2 Neutrosophic Fuzzy-Weighted Zero-Inconsistency" (T2NN-FWZIC) technique is formulated for weighting the evaluation criteria of RAN architectures. Second, another new method, namely, "Type-2 Neutrosophic Fuzzy Decision by Opinion Score Method" (T2NN-FDOSM), was formulated to select the optimal RAN architecture using the obtained weights. The weighting results by T2NN-FWZIC for the (n = 25) evaluation criteria revealed that (C 21 latency and C 22 reliability) as the most important criteria, with 0.06 value for each as opposed to (C 15 Data Processing) as the lowest weighted criteria with 0.0186 value. As for T2NN-FDOSM, a total of four 5G-RAN architectures were evaluated, including "virtualized cloud RAN" coming as the optimal one, followed by "fog RAN," "cloud RAN," and finally "heterogeneous cloud RAN." The results were confirmed by carrying out a sensitivity analysis. The outcome of this study can be used to assist future 5G-RAN developments according to business needs and to establish an assessment platform for 5G technology in different domains and applications. [ABSTRACT FROM AUTHOR]

This study presents a novel dynamic localisation-based decision (DLBD) with fuzzy weighting with zero inconsistency (FWZIC) under a probabilistic single-valued neutrosophic hesitant fuzzy set (PSVNHFS) environment to benchmark Hybrid Multi Deep Transfer and Machine Learning (HMDTML) models. The novel DLBD method is proposed to generate a dynamic localisation decision matrix based on the upper and lower boundaries and the length of the scale. The superiority of DLBD derives from its ability to manage dynamic changes with boundary value consequences. In addition, the utilization of PSVNHFS in conjunction with DLBD and FWZIC has proven to effectively address the challenges posed by vagueness, uncertainty and hesitancy in the benchmarking procedure. The proposed methodology consists of three primary three steps: i) the adaptation of 48 HMDTML models, including 4 deep transfer learning models and 12 machine learning models trained on a dataset of 936 chest X-ray images obtained from both COVID-19 patients and individuals without the disease. Then, these models were evaluated based on seven evaluation criteria, and a decision matrix was proposed. ii) The development of a PSVNH–FWZIC to assign weights to the evaluation criteria. iii) The formulation of a PSVNH–DLBD for the purpose of benchmarking HMDTML models. Results of the PSVNH–FWZIC revealed that AUC and time were the most important evaluation criteria, while precision was the least important. Furthermore, the results from PSVNH–DLBD, reveal that Model M24 (Painters-Decision Tree) earned the highest rank when λ = 2 , 3 , 4 , 5 a n d 6 , followed by Model M25 (SqueezeNet-AdaBoost) and Model M34 (DeepLoc-kNN), while Model M39 (DeepLoc-SVM) had the lowest rank (rank = 48) across all λ values. The proposed method underwent sensitivity and comparison analyses to confirm its reliability and robustness. [ABSTRACT FROM AUTHOR]

The digital twin (DT) represents a powerful tool for advancing construction industry to provide a cyber–physical integration that enables real-time monitoring of assets and activities and facilitates decision-making. Due to the inherent characteristics of the construction industry and the diverse possibilities with DT, proliferation of building digital twin (BDT) necessitates a comprehensive comprehension of its evolution and the creation of roadmaps. This paper aims to contribute to the formalization and standardization of BDT. It designs a novel assessment framework for the overall maturity measurement of existing BDT projects. The developed BDT maturity model incorporates a collective opinion generation paradigm based on a fairness-aware multiobjective optimization model to provide an expert-based evaluation system for evaluating the maturity of BDT projects. The effectiveness and feasibility of the proposed framework have been validated through a case study of an experimental BDT initiative. This paper establishes a generalizable framework for BDT maturity assessment that can offer insights into BDT maturity standards to construction practitioners to create effective strategies for the diffusion, development, and maturation of BDT. [ABSTRACT FROM AUTHOR]

The growth potential of a career reflects its future prospects and is an important consideration for individuals and organizations when career planning. There is still a lack of quantitative assessment tools for growth potential of careers. In this study, considering the key role of human capital in human resource management, as well as the excellent performance of complex network and machine learning in big data analysis and prediction, a career growth potential assessment model with human capital ensemble is proposed through human capital-based occupational mobility network and ensemble learning. First, an occupational mobility network is constructed based on online professional dataset to associate occupations with each other. Then, five dimensions of human capital measurements are designed to quantify human capital in terms of education, experience, social capital, occupational size, and concentration. These are then combined with the occupational mobility network to create a new network that depicts human capital flows among occupations. Finally, an ensemble framework for assessing career growth potential is constructed to integrate multidimensional human capital information in the network and obtain quantitative scores of growth potential. This study is the original attempt to adopt a data-driven idea and an intelligent approach to understand career growth potential. The experimental results show that it also makes a useful exploration for modeling human capital flows and intelligent assessment of career prospects. [ABSTRACT FROM AUTHOR]

Metaverse is defined as a fictional universe that could serve as a simulation environment of reality. Beginning in the past with games, it becomes increasingly integrated into human life as time passes. Metaverse usage is inevitable in every aspect of life. One of its potential application areas could be urban transportation. A novel fuzzy trigonometric based on the combination of the Full Consistency Method (FUCOM) and Combined Compromise Solution (CoCoSo) is proposed to rank three alternatives with twelve criteria under four major aspects: managerial, safety, user, and urban mobility. In the first stage, fuzzy FUCOM methods are used to calculate the weights of the criteria. In the second stage, the fuzzy trigonometric based CoCoSo method is applied to evaluate and rank the alternatives. The proposed model enables the nonlinear processing of complex and uncertain information using fuzzy trigonometric functions. The findings demonstrate focusing on a particular age group can make it easier to integrate the metaverse with urban transportation. The findings of this study have the potential to serve as a guide for decision-makers. The metaverse-based applications could be started by policymakers, which is a promising opportunity with potential boundaries beyond human comprehension making this statement weaker. [ABSTRACT FROM AUTHOR]

A broad-deep fusion network-based fuzzy emotional inference model with personal information (BDFEI) is proposed for emotional intention understanding in human–robot interaction. It aims to understand students' intentions in the university teaching scene. Initially, we employ convolution and maximum pooling for feature extraction. Subsequently, we apply the ridge regression algorithm for emotional behavior recognition, which effectively mitigates the impact of complex network structures and slow network updates often associated with deep learning. Moreover, we utilize multivariate analysis of variance to identify the key personal information factors influencing intentions and calculate their influence coefficients. Finally, a fuzzy inference method is employed to gain a comprehensive understanding of intentions. Our experimental results demonstrate the effectiveness of the BDFEI model. When compared to existing models, namely FDNNSA, ResNet-101+GFK, and HCFS, the BDFEI model achieved superior accuracy on the FABO database, surpassing them by 12.21%, 1.89%, and 0.78%, respectively. Furthermore, our self-built database experiments yielded an impressive 82.00% accuracy in intention understanding, confirming the efficacy of our emotional intention inference model. [ABSTRACT FROM AUTHOR]