Search

Nowadays, there has been a rapidly increasing number of scientific submissions in multiple research domains. A large number of journals have various acceptance rates, impact factors, and rankings in different publishers. It becomes time-consuming for many researchers to select the most suitable journal to submit their work with the highest acceptance rate. A paper submission recommendation system is more critical for the research community and publishers as it gives scientists another support to complete their submission conveniently. This paper investigates the submission recommendation system for two main research topics: computer science and applied mathematics. Unlike the previous works (Wang et al., 2018; Son et al., 2020) that extract TF–IDF and statistical features as well as utilize numerous machine learning algorithms (logistics regression and multiple perceptrons) for building the recommendation engine, we present an efficient paper submission recommendation algorithm by using different bidirectional transformer encoders and the Mixture of Transformer Encoders technique. We compare the performance between our methodology and other approaches by one dataset from Wang et al. (2018) with 14012 papers in computer science and another dataset collected by us with 223,782 articles in 178 Springer applied mathematics journals in terms of top K accuracy (K = 1 , 3 , 5 , 10). The experimental results show that our proposed method extensively outperforms other state-of-the-art techniques with a significant margin in all top K accuracy for both two datasets. We publish all datasets collected and our implementation codes for further references. 1 1 https://github.com/BinhMisfit/PSRMTE. • Bidirectional transformer encoders can improve the performance of the paper submission recommendation system. • The Mixture of Transformer Encoders framework shows the efficiency in the paper submission recommendation problem. • Proposed techniques can surpass other recent techniques on two datasets related. [ABSTRACT FROM AUTHOR]

Multiple Instance Learning (MIL) is a learning paradigm, where training instances are arranged in sets, called bags, and only bag-level labels are available during training. This learning paradigm has been successfully applied in various real-world scenarios, including medical image analysis, object detection, image classification, drug activity prediction, and many others. This survey paper presents a comprehensive analysis of MIL, highlighting its significance, recent advancements, methodologies, applications, and evolving trends across diverse domains. The survey begins by explaining the core principles that form the basis of MIL and how it differs from traditional learning approaches. This sets the foundation for comprehending the distinct challenges and techniques of solving MIL problems. Next, we discuss how supervised learning algorithms are tailored to support MIL and combine this discussion with a review of seminal MIL algorithms as well as the latest innovations that incorporate neural networks, deep learning architectures, and attention techniques. This comprehensive analysis helps to understand the strengths, limitations, and adaptability of these methods across diverse data modalities, complexities, and applications. In summary, this survey paper provides an essential resource for researchers, practitioners, and enthusiasts seeking a comprehensive understanding of Multiple Instance Learning. It covers foundational concepts, traditional methods, recent advancements, and future directions. By providing a holistic view of MIL's dynamic landscape, this paper aims to inspire further innovation and exploration in this ever-evolving field. • A survey on the current state of the Multiple Instance Learning. • We provide applications of Multiple instance learning in various domains. • We discuss how existing supervised learning algorithms are modified to support MIL. • Publicly available datasets and open research challenges in MIL are discussed. [ABSTRACT FROM AUTHOR]

This paper proposes physics-informed meta-learning-based surrogate modeling (PI-MLSM), a novel approach that combines meta-learning and physics-informed deep learning to train surrogate models with limited labeled data. PI-MLSM consists of two stages: meta-learning and physics-informed task adaptation. The proposed approach is demonstrated to outperform other methods in four numerical examples while reducing errors in prediction and reliability analysis, exhibiting robustness, and requiring less labeled data during optimization. Moreover, compared to other approaches, the proposed approach exhibits better performance in solving out-of-distribution tasks. Although this paper acknowledges certain limitations and challenges, such as the subjective nature of physical information, it highlights the key contributions of PI-MLSM, including its effectiveness in solving a wide range of tasks and its ability in handling situations wherein physical laws are not explicitly known. Overall, PI-MLSM demonstrates potential as a powerful and versatile approach for surrogate modeling. • A proposed approach uses meta-learning and PIDL to train surrogates with small data. • A model-agnostic meta-learning model is introduced to learn surrogate model weights. • A PIDL approach is introduced to adapt to target tasks with meta-learned weights. • Our approach outperformed other methods in numerical examples given limited data. • Proposed approach showed robustness in solving out-of-distribution tasks. [ABSTRACT FROM AUTHOR]

In this paper the performance of different Machine Learning and Deep Learning approaches is evaluated in problems related to green mobility in big cities. Specifically, the forecasting of bike sharing demand in Madrid and Barcelona (Spain) is approached, for different prediction time-horizons, and also a problem of cable car demand forecasting in Madrid city. An important number of predictive variables are considered, which are grouped into four different sets (categorical/calendrical, persistence-based, meteorological and, as a novelty of the paper, information about analogue past instances), whose relevance is studied for all cases. A feature selection mechanism is also incorporated in order to improve the prediction accuracy of the proposed algorithms. A total of 12 different multivariate regression techniques are implemented, covering from Machine Learning methods to time-series Deep Learning approaches. Excellent results in all the prediction problems approached are reported. Finally, the consequences of obtaining accurate prediction in these three problem of green mobility in big cities are discussed. In addition, it is studied how the results could be exported to other similar cases in more general urban mobility studies. Novelties of the work include: (1) Addressing the forecast problem of passenger flow on a cable car using ML and DL multivariate techniques; (2) using the demand of analogous past instances as an additional feature to solve the demand prediction problems; and (3) the extraction of global conclusions about feature relevance when addressing a demand forecasting problem in green mobility. [ABSTRACT FROM AUTHOR]

Intelligent Fault-Tolerant Control (IFTC) refers to the applications of machine learning algorithms for fault diagnosis and design of Fault-Tolerant Control (FTC). The overall goal of the FTC is to accommodate defects in the system components while they are in use and maintain stability with little to no performance reduction. These systems are crucial for mission-critical and safety-related applications where the safety of people is at stake and service continuity is crucial. In this review paper, a systematic study has been done for the development of FTC with machine learning, deep learning, and transfer learning algorithms. The challenges and limitations faced with their possible solutions through machine learning theories for the IFTC model are lined up. This paper guides researchers on the different possible types of machine learning algorithms and their advanced forms like deep learning and transfer learning. The differences among these are highlighted by the challenges and limitations of each. The paper is significant such that most of the important literature references from the Scopus database particularly related to important electrical and mechanical industrial problems have been discussed to guide the researchers who want to apply IFTC for specific industrial problems, being the research gap. Finally, future research directions for the development of IFTC are highlighted. [ABSTRACT FROM AUTHOR]

Marine ecosystems and their fish habitats are becoming increasingly important due to their integral role in providing a valuable food source and conservation outcomes. Due to their remote and difficult to access nature, marine environments and fish habitats are often monitored using underwater cameras to record videos and images for understanding fish life and ecology, as well as for preserve the environment. There are currently many permanent underwater camera systems deployed at different places around the globe. In addition, there exists numerous studies that use temporary cameras to survey fish habitats. These cameras generate a massive volume of digital data, which cannot be efficiently analysed by current manual processing methods, which involve a human observer. Deep Learning (DL) is a cutting-edge Artificial Intelligence (AI) technology that has demonstrated unprecedented performance in analysing visual data. Despite its application to a myriad of domains, its use in underwater fish habitat monitoring remains under explored. In this paper, we provide a tutorial that covers the key concepts of DL, which help the reader grasp a high-level understanding of how DL works. The tutorial also explains a step-by-step procedure on how DL algorithms should be developed for challenging applications such as underwater fish monitoring. In addition, we provide a comprehensive survey of key deep learning techniques for fish habitat monitoring including classification, counting, localisation, and segmentation. Furthermore, we survey publicly available underwater fish datasets, and compare various DL techniques in the underwater fish monitoring domains. We also discuss some challenges and opportunities in the emerging field of deep learning for fish habitat processing. This paper is written to serve as a tutorial for marine scientists who would like to grasp a high-level understanding of DL, develop it for their applications by following our step-by-step tutorial, and see how it is evolving to facilitate their research efforts. At the same time, it is suitable for computer scientists who would like to survey state-of-the-art DL-based methodologies for fish habitat monitoring. • Marine environments and fish habitats are widely surveyed using underwater cameras. • Deep learning (DL) offers an effective solution for processing underwater videos. • We write a tutorial on the key concepts and steps of DL for underwater monitoring. • We provide a survey of key DL-based visual fish monitoring methods/datasets. • We shed light on challenges/opportunities of DL for underwater fish monitoring. [ABSTRACT FROM AUTHOR]

In this paper we present SurvLIMEpy, an open-source Python package that implements the SurvLIME algorithm. This method allows to compute local feature importance for machine learning algorithms designed for modelling Survival Analysis data. The presented implementation uses a matrix-wise formulation, which allows to speed up the execution time. Additionally, SurvLIMEpy assists the user with visualisation tools to better understand the result of the algorithm. The package supports a wide variety of survival models, from the Cox Proportional Hazards Model to deep learning models such as DeepHit or DeepSurv. Two types of experiments are presented in this paper. First, by means of simulated data, we study the ability of the algorithm to capture the importance of the features. Second, we use three open source survival datasets together with a set of survival algorithms in order to demonstrate how SurvLIMEpy behaves when applied to different models. • Python package implementing SurvLIME algorithm for computing feature importance. • It supports a wide variety of survival models, including deep learning models. • Fast and efficient implementation due to matrix-wise optimisation problems. • An open-source implementation available on GitHub. • Stable release provided to PyPI. [ABSTRACT FROM AUTHOR]

Epilepsy is one of the most paramount neurological diseases, affecting about 1% of the world's population. Seizure detection and classification are difficult tasks and are ongoing challenges in biomedical signal processing to enhance medical diagnosis. This paper presents and highlights the unique frequency and amplitude information found within multiple seizure types, including their morphologies, to aid the development of future seizure classification algorithms. Whilst many published works in the literature have reported on seizure detection using electroencephalogram (EEG), there has yet to be an exhaustive review detailing multi-seizure type classification using EEG. Therefore, this paper also includes a detailed review of multi-seizure type classification performance based on the Temple University Hospital Seizure Corpus (TUSZ) dataset for focal and generalised classification, and multi-seizure type classification. Deep learning techniques have a higher overall average performance for focal and generalised classification compared to machine learning techniques, whereas hybrid deep learning approaches have the highest overall average performance for multi-seizure type classification. Finally, this paper also highlights the limitations of the TUSZ dataset and suggests some future work, including the curation of a standardised training and testing dataset from the TUSZ that would allow a proper comparison of classification methods and spur advancement in the field. [ABSTRACT FROM AUTHOR]

Automatic fault diagnosis in many systems is performed via the analysis of vibration data in the time or frequency domain. In the literature, many approaches for extracting featured from signals for fault diagnosis have been proposed. In this paper, we apply a variety of transform functions including Fourier, Wavelet, etc. to extract features from the vibration data, which includes the statistical features and some features automatically extracted via Convolutional Neural Networks (CNNs). For each of these feature extraction approaches, a learning algorithm is trained to diagnose the faults and their results are aggregated in an ensemble machine learning algorithm. The weights of the base learner algorithms are optimized via an evolutionary algorithm to achieve the best-weighted voting scheme. The architecture of CNNs has a significant effect on the performance of the algorithm, thus, in this paper, an evolutionary algorithm is proposed to find the best architecture for CNNs in fault diagnosis. The CNNs are trained via gradient descent algorithms which suffer from getting stuck in local optima. To manage this, we propose an evolutionary algorithm that benefits from the speed of gradient descent and the global search of evolutionary algorithms. The proposed algorithm is tested on a number of benchmark problems and the experimental results are presented. • An ensemble algorithm is proposed for fault diagnosis. • Different feature extractors are exploited. • An evolutionary training algorithm is proposed. [ABSTRACT FROM AUTHOR]

Fake news refuting is an essential part of online social networks (OSNs) management. Identifying the potential voluntary fake news refuters who can spontaneously repost the fake news refutation messages on the OSNs, and intentionally exposing such messages to them can contribute a lot to fake news regulation. Therefore, to fully utilize available user features and enhance fake news refuter identification, the paper proposes a novel paradigm for refuter classification problem, which leverages the pre-trained techniques and mixture of expert system to unify the structured and unstructured data. Specifically, the unstructured text records are first understood by the pre-trained ERNIE2 model, and then feed into Chi-Merge grounded Convolutional Neural Networks, which are expert classifiers in our framework, along with the structured profile data for further transformation. Besides, features extracted from text data are used for determining weights for multiple expert classifiers. Experiments on near 60 k real-world Weibo user data prove that our framework outperforms the large-scale pre-trained deep learning methods and traditional machine learning classification models. This paper provides a novel perspective of fake news refutation from the refuter's side and sheds light on artificial intelligence-driven OSNs management. [ABSTRACT FROM AUTHOR]

Parkinson's disease (PD) is a neurodegenerative disorder that produces both motor and non-motor complications, degrading the quality of life of PD patients. Over the past two decades, the use of wearable devices in combination with machine learning algorithms has provided promising methods for more objective and continuous monitoring of PD. Recent advances in artificial intelligence have provided new methods and algorithms for data analysis, such as deep learning (DL). The aim of this article is to provide a comprehensive review of current applications where DL algorithms are employed for the assessment of motor and non-motor manifestations (NMM) using data collected via wearable sensors. This paper provides the reader with a summary of the current applications of DL and wearable devices for the diagnosis, prognosis, and monitoring of PD, in the hope of improving the adoption, applicability, and impact of both technologies as support tools. Following PRISMA (Systematic Reviews and Meta-Analyses) guidelines, sixty-nine studies were selected and analyzed. For each study, information on sample size, sensor configuration, DL approaches, validation methods and results according to the specific symptom under study were extracted and summarized. Furthermore, quality assessment was conducted according to the Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD) method. The majority of studies (74%) were published within the last three years, demonstrating the increasing focus on wearable technology and DL approaches for PD assessment. However, most papers focused on monitoring (59%) and computer-assisted diagnosis (37%), while few papers attempted to predict treatment response. Motor symptoms (86%) were treated much more frequently than NMM (14%). Inertial sensors were the most commonly used technology, followed by force sensors and microphones. Finally, convolutional neural networks (52%) were preferred to other DL approaches, while extracted features (38%) and raw data (37%) were similarly used as input for DL models. The results of this review highlight several challenges related to the use of wearable technology and DL methods in the assessment of PD, despite the advantages this technology could bring in the development and implementation of automated systems for PD assessment. • Deep learning and wearable technology are used in diagnosis and monitoring. • Growing interest in the use of deep learning and wearables in Parkinson's disease. • The adoption of deep learning in Parkinson's disease is currently under development. • Technical and methodological challenges difficult the technology adoption. [ABSTRACT FROM AUTHOR]

Artificial intelligence (AI) technologies are increasingly applied to traffic flow prediction (TFP) to enhance prediction accuracy. This study utilizes bibliometric methods and network analysis measures to gain insights into the research status, development process, opportunities, and challenges of AI-based TFP research based on the literature data retrieved from the Web of Science core collection. The study first conducts basic statistical analysis of all papers. Subsequently, cooperation network analysis is conducted to identify the most productive countries/territories, institutions, and authors, the cooperative relationships, and the formed research communities. Co-citation network analysis is then employed to identify publications that have made outstanding contributions to the AI-based TFP field. Finally, the main path analysis of the paper citation network is used to analyze the knowledge diffusion process, while the keyword co-occurrence analysis is conducted to reveal the evolution characteristics of the research topics. Based on the bibliometric analysis results, we gain insights into the opportunities and challenges in this field from the perspectives of data, models, and applications, and provide pertinent suggestions for future research. Overall, this study can assist researchers in capturing the state-of-the-art and research directions in AI-based TFP. [ABSTRACT FROM AUTHOR]

This paper proposes a robust algorithm using two convolutional neural networks (CNN) to find a lost gamma source in a room despite various unpredictable obstacles. This algorithm is an improved and complementary method to our previous research, which does not limit the use of the algorithm to a specific geometry, and the presence of obstacles does not reduce the performance and inability of the method. In our previous research, an algorithm including a CNN was used to estimate the location of the lost gamma source in known geometry. In this paper, by adding another CNN to the previous algorithm, the algorithm can be used in environments with unknown geometry. The environmental simulations and data generation are performed using the Geant4 simulation program. The method has been tested in a designed scenario to check its effectiveness. The results show that by using these two CNNs, an agent can intelligently find the gamma source place in a room with multiple obstacles without human intervention with 95 percent accuracy. [ABSTRACT FROM AUTHOR]

Shoeprint reconstruction is essential in forensic science, but it is also challenging due to various inconsistencies in the patterns, textures, sizes, abrasions, etc. of shoeprints. The computational reconstruction of sharper and more complete shoeprints is conventionally conducted using handcrafted features, and it often requires human intervention. Prior studies using end-to-end machine learning approaches are limited in number and have not achieved a high level of performance. In this paper, we propose a model named ShoeRec, which employs variational autoencoder (VAE) as a component in a U-Net-like architecture to reconstruct missing regions and borders in shoeprint images. ShoeRec incorporates skip connections to preserve key patterns and employs VAE in the bottleneck to facilitate the reconstruction of desired shoeprints with the restoration of detail as perceived by humans. As a U-Net, the model skips the contextual information from the encoder to the decoder, and the compressed features in the latent space via VAE optimize the probabilistic distribution for reconstructing complete shoeprints. The reconstruction operation is automatically tuned according to the objective function, so as to reduce the structural correlation between the original and projected shoeprint and restore the absent information in the desired shoeprints. To the best of our knowledge, ShoeRec is the first deep learning infusion model that specializes in shoeprint reconstruction. The shoeprints reconstructed by ShoeRec have a close match with the originals in terms of both structures and patterns, and ShoeRec outperforms state-of-the-art generative models in human evaluation. • An end-to-end deep learning model focuses on shoeprint reconstruction for forensics. • Reconstruction of existing patterns together with the generation of missing regions. • Missing regions reconstruction is specifically designed based on the loss function. • The proposed model performs well for noisy, blurry, and partial input images. • Model outperforms state-of-the-art generative models in terms of human evaluation. [ABSTRACT FROM AUTHOR]

• The paper explores AI decision-making to enhance transparency of CIN diagnosis using colposcopy images. • The study presents interpretable insights for clinical validation using different XAI methods. • Saliency method showed best result, using gradients to reveal likely malignancy traits. • Expert Clinic Validation of results conducted, along with statistical analyses. Although artificial intelligence techniques have performed well in analysing medical images, it is essential to consider the degree of understanding or interpretation in evaluations. This is especially important when diagnostic support is involved, as the trustworthiness of machine learning predictions needs to be clarified and evaluated. For deep learning algorithms to be translated into clinical practice, they must be made less opaque. To examine the classification of cervical dysplasia-diagnosed patients, we will explore explainable artificial intelligence (XAI) approaches, by implementing seven different interpretation techniques (Saliency, XRAI, Integrated Gradients, Smooth Gradients, Smooth Integrated Gradients, Grad-CAM, Smoothgrad-CAM). These XAI techniques shall be explored in order to assess the performance of four Convolutional Neural Network models (AlexNet, VGG16, MobileNet, and InceptionNet). Our results show that saliency qualities most closely match expert annotations, and there is a moderate to strong association between a model's sensitivity and the agreement between humans and computers. We also discovered a relevant relationship between expert insights and computational learning. This highlights the potential role of human expertise in influencing effective computer learning. [ABSTRACT FROM AUTHOR]

Correctly predicting up trends, sideways trends and down trends for stock returns is vitally important in financial market. Group penalized trinomial logit dynamic models can not only provide three-class label information and three-class probability estimations, but also enhance three-class prediction performance by shrinking group coefficients to bypass multi-collinearity and over-fitting. In this paper we propose G-LASSO/G-SCAD/G-MCP penalized trinomial logit dynamic models with 24 technical indicators to predict up trends, sideways trends and down trends for stock returns, develop group coordinate descent algorithm (GCD) to complete group selection and group estimation simultaneously, establish the relative optimal Bayes classifier to identify class labels, introduce three-class confusion matrix and hypervolume under the ROC manifold (HUM) to assess the three-class prediction performance to 15 methods. Experiment results show that G-LASSO/G-SCAD/G-MCP penalized trinomial logit dynamic models predict better than LASSO/SCAD/MCP penalized trinomial logit dynamic models, 6 deep learning models and 3 machine learning models in terms of Accuracy and HUM. In particular, the highest prediction Accuracy (the average Accuracy) from G-LASSO penalized method outperforms 3-Layer Long Short-Term Memory (LSTM) and Random Forest (RF) for 4.76% and 10.29% (8.16% and 11.78%), respectively. Moreover, compared with LASSO/SCAD/MCP penalized methods, G-LASSO/G-SCAD/G-MCP penalized methods enhance 5.88% Accuracy and only require 27.29% Predicted Time. The proposed G-LASSO/G-SCAD/G-MCP penalized trinomial logit dynamic models can not only be extended as more general three-class prediction method by replacing 24 technical indicators by some factors influencing three-class response variable, but also be directly extended as more efficient multi-category prediction method by replacing three-class response variable by multi-category response variable. • Propose the new group penalized methods to predict stock return movement direction. • Develop a new GCD algorithm for G-LASSO/G-SCAD/G-MCP penalized trinomial logit model. • Obtain the new rapidly convergent G-LASSO/G-SCAD/G-MCP estimator. • Introduce a few measures to assess three-class prediction performance to 15 methods. • G-LASSO/G-SCAD/G-MCP penalized methods predict better than the other 12 methods. [ABSTRACT FROM AUTHOR]

Active learning is focused on minimizing the effort required to obtain labeled data by iteratively choosing fresh data samples for training a machine learning model. One of the primary challenges in active learning involves the selection of the most informative instances for labeling by an annotation oracle at each iteration. A viable approach is to develop an active learning strategy that aligns with the performance of a meta-learning model. This strategy evaluates the quality of previously selected instances and subsequently trains a machine learning model to predict the quality of instances to be labeled in the current iteration. This paper introduces a novel approach to learning for active learning, wherein instances are chosen for labeling based on their potential to induce the most substantial change in the current classifier. We explore various strategies for assessing the significance of an instance, taking into account variations in the learning gradient of the classification model. Our approach can be applied to any classifier that can be trained using gradient descent optimization. Here, we present a formulation that leverages a deep neural network model, which has not been extensively explored in existing learning-to-active-learn methodologies. Through experimental validation, our approach demonstrates promising results, especially in scenarios where there are limited initially labeled instances, particularly when the number of labeled instances per class is extremely limited. • A new meta-active learning strategy for deep neural network models. • The instance selection problem is modeled as a regression problem. • Four strategies of instance importance scoring by considering gradient variations. • Experiments have shown good results when few initially labeled objects are available. • A dimensionality reduction step for the instances improves the performances. [ABSTRACT FROM AUTHOR]

Traffic accidents are usually unique events with unpredictable geographical and temporal dimensions; thus, accident injury severity level (INJ-SL) analysis presents formidable categorization and data stability problems. Classical statistical models are limited in their ability to correctly model INJ-SL, whilst sophisticated machine learning approaches do not appear to have any equations to prioritize/analyze multiple contributing factors to forecast accidents accompanying INJ-SLs. In addition, the intercorrelations between the input variables may render the conclusions of a formal sensitivity analysis incorrectly. Rear-end collisions are the most common form of traffic accidents; consequently, their linked INJ-SL requires more research. This paper provides a complex technique based on a deep learning paradigm paired with different indicators of Global Sensitivity Analysis to address all of these concerns. Unlike existing neural network designs, this technique presents a deep residual neural network structure that employs residual shortcuts (i.e., connections). The connections enable the DRNNs to sidestep a few levels of the deep network architecture, evading the regular training with high accuracy issues. Using the trained DRNNs model, a Latin Hypercube sampling simulation was undertaken to determine each explanatory component's influence on the resulting INJ-SL. The latest available data from 2011 to 2018 is used to assess all rear-end collisions in North Carolina. A comparison was made between the performance of two different schemes of data categorization using a set of global sensitivity metrics. It was determined that the devised technique overcame the data heterogeneity problems to achieve an accuracy of 87%. In addition, the proposed sensitivity analysis identified the most relevant factors associated with INJ-SL rear-end collisions. [ABSTRACT FROM AUTHOR]

Text summarization is a widely-researched problem among scholars in the field of natural language processing. Multiple techniques have been proposed to help tackle this problem, yet some of these methodologies may still exhibit limitations such as the requirements for large training datasets, which might not always be possible, but more importantly, the lack of interpretability or transparency of the model. In this paper, we propose using meta-learning algorithm to train a deep learning model for extractive text summarization and then using various explanatory techniques such as SHAP (Shapley, 1953), linear regression (Lederer, 2022), decision trees (Fürnkranz, 2010), and input modification to gain insights into the model's decision making process. To evaluate the effectiveness of our approach, we will compare it to other popular natural language processing models like BERT (Miller, 2019) or XLNET (Yang et al., 2020) using the ROUGE metrics (Lin, 2004). Overall, our proposed approach provides a promising solution to the limitations of existing methods and a framework for improving the explainability of deep learning models in other natural language processing tasks. • Bi-LSTM and Meta-learning produce better results compared to other popular models. • Meta-learning can be utilized to improve performance in data-scarce scenarios. • Use SHAP, regression, decision tree, and text mutation to enhance interpretability. [ABSTRACT FROM AUTHOR]

Highlights • The paper combines domain knowledge (DK) with machine learning (ML). • Patterns of logical analysis of data (LAD) are used by experts to enrich the DK. • The LAD patterns identify fault causes that are not represented in the DK. • The identified causes are used to enrich the fault trees in industrial plants. Abstract Experts in industrial processes rely on domain knowledge (DK) repositories to identify the causes of abnormal situations in order to make appropriate decisions that mitigate the negative effects of such events. These DK repositories need to be enriched and updated continuously as different unexpected events occur. A common causality analysis method in DK repositories is the fault tree analysis (FTA). The major limitation of updating a fault tree is that it requires in-depth system knowledge, which involves a high level of human experience. Data exploitation based on machine learning (ML) can address this limitation by deeply analyzing process historical data to discover hidden phenomena that are difficult for human experts to identify and to analyze. This paper proposes an innovative methodology that combines domain knowledge, in the form of FTA, with additional knowledge extracted by a descriptive ML method called logical analysis of data (LAD). More specifically, LAD is a classification method, which provides as a by-product a set of interpretable rules (patterns) explaining the classification results. The patterns extracted from historical data represent an important and complementary source of knowledge that provides experts with insights and allows them to better understand the process operations. The objective of using these patterns in the proposed methodology is to provide automatic enrichment and updating of existing fault trees in order to achieve accurate fault detection and diagnosis (FDD) in industrial processes. The proposed methodology is demonstrated using fault trees constructed for two different systems in the process industry. The fault tree for each system was updated successfully with minimal effort from process experts. [ABSTRACT FROM AUTHOR]

For the purpose of managing financial risk and making investment decisions, interval stock price forecasting is essential. Currently, decomposition integration frameworks are widely used in point-valued stock price forecasting studies, mainly focusing on mining internal information. However, point forecasts are difficult to adequately capture price uncertainty and may suffer from loss of volatile information. Therefore, this paper proposes an enhanced interval-valued decomposition integration model for stock price prediction based on comprehensive feature extraction and optimized deep learning. Firstly, the interval variational modal decomposition with feedback mechanism (FIVMD) is proposed to extract internal features and can decompose interval values into interval trend and residual. FIVMD not only solves the interval decomposition challenge, but also helps to improve the internal feature extraction capability. Secondly, while considering the influencing factors more comprehensively, appropriate feature selection and compression techniques can effectively achieve external feature extraction, obtain the best influencing factors, and improve the modeling capability of high-dimensional data. Finally, the final prediction results are obtained by modeling the interval trend and residuals separately through the optimization algorithm and deep learning model to improve the prediction accuracy. The results of the empirical analysis reveal that the proposed interval decomposition integrated model has the smallest of the three evaluation metrics, where the values of interval mean average percentage errors (IMAPE) are 1.8188%, 1.1244%, 1.9001%, and 2.1542% respectively. This shows that the model is significantly more accurate and stable than the other comparative models, and it is a successful model for predicting interval-valued stock prices. [ABSTRACT FROM AUTHOR]

In 2016, Google proposed Federated Learning (FL) as a novel paradigm to train Machine Learning (ML) models across the participants of a federation while preserving data privacy. Since its birth, Centralized FL (CFL) has been the most used approach, where a central entity aggregates participants' models to create a global one. However, CFL presents limitations such as communication bottlenecks, single point of failure, and reliance on a central server. Decentralized Federated Learning (DFL) addresses these issues by enabling decentralized model aggregation and minimizing dependency on a central entity. Despite these advances, current platforms training DFL models struggle with key issues such as managing heterogeneous federation network topologies, adapting the FL process to virtualized or physical deployments, and using a limited number of metrics to evaluate different federation scenarios for efficient implementation. To overcome these challenges, this paper presents Fedstellar, a novel platform designed to train FL models in a decentralized, semi-decentralized, and centralized fashion across diverse federations of physical or virtualized devices. Fedstellar allows users to create federations by customizing parameters like the number and type of devices training FL models, the network topology connecting them, the machine and deep learning algorithms, or the datasets of each participant, among others. Additionally, it offers real-time monitoring of model and network performance. The Fedstellar implementation encompasses a web application with an interactive graphical interface, a controller for deploying federations of nodes using physical or virtual devices, and a core deployed on each device, which provides the logic needed to train, aggregate, and communicate in the network. The effectiveness of the platform has been demonstrated in two scenarios: a physical deployment involving single-board devices such as Raspberry Pis for detecting cyberattacks and a virtualized deployment comparing various FL approaches in a controlled environment using MNIST and CIFAR-10 datasets. In both scenarios, Fedstellar demonstrated consistent performance and adaptability, achieving F 1 s c o r e s of 91%, 98%, and 91.2% using DFL for detecting cyberattacks and classifying MNIST and CIFAR-10, respectively, reducing training time by 32% compared to centralized approaches. • Open-source platform for training FL models in a decentralized fashion. • Support for diverse FL architectures, network topologies, datasets, and ML/DL models. • User-friendly setup and efficient monitoring of FL processes. • Successful deployment across diverse physical and virtualized devices. • Promising performance in tasks like cyberattack detection and image classification. [ABSTRACT FROM AUTHOR]

Deep learning has emerged as a powerful tool in various domains, revolutionising machine learning research. However, one persistent challenge is the scarcity of labelled training data, which hampers the performance and generalisation of deep learning models. To address this limitation, researchers have developed innovative methods to overcome data scarcity and enhance deep model learning capabilities. Two prevalent techniques that have gained significant attention are transfer learning and self-supervised learning. Transfer learning leverages knowledge learned from pre-training on a large-scale dataset, such as ImageNet, and applies it to a target task with limited labelled data. This approach allows models to benefit from the learned representations and effectively transfer knowledge to new tasks, resulting in improved learning performance and generalisation. On the other hand, self-supervised learning focuses on training models using pretext tasks that do not require manual annotation, allowing them to learn valuable representations from large amounts of unlabelled data. These learned representations can then be fine-tuned for downstream tasks, mitigating the need for extensive labelled data. In recent years, transfer and self-supervised learning have found applications in various fields, including medical image processing, video recognition, and natural language processing. These approaches have demonstrated remarkable achievements, enabling breakthroughs in areas such as disease diagnosis, object recognition, and language understanding. However, while these methods offer numerous advantages, they also have limitations. For example, transfer learning may face domain mismatch issues between the pre-training and target domains, while self-supervised learning requires careful design of pretext tasks to ensure meaningful representations. This review paper explores the recent applications of these pre-training methods in various fields within the past three years. It delves into the advantages and limitations of each approach, assesses the performance of models employing these techniques, and identifies potential directions for future research. By providing a comprehensive review of current pre-training methods, this article offers guidance for selecting the best technique for specific deep learning applications to address the data scarcity issue. • Differences between Transfer Learning and Self-Supervised Learning. • Applications of Transfer Learning and Self-Supervised Learning. • Advantages and Limitations of Transfer Learning and Self-Supervised Learning. • Recommendations for the Use of Each Approach in Specific Applications. [ABSTRACT FROM AUTHOR]

In recent years, the enormous development of Machine Learning, especially Deep Learning, has led to the widespread adoption of Artificial Intelligence (AI) systems in a large variety of contexts. Many of these systems provide excellent results but act as black-boxes. This can be accepted in various contexts, but there are others (e.g., medical ones) where a result returned by a system cannot be accepted without an explanation on how it was obtained. Explainable AI (XAI) is an area of AI well suited to explain the behavior of AI systems that act as black-boxes. In this paper, we propose a model-agnostic XAI framework to explain the behavior of classifiers. Our framework is based on network theory; thus, it is able to make use of the enormous amount of results that researchers in this area have discovered over time. Being network-based, our framework is completely different from the other model-agnostic XAI approaches. Furthermore, it is parameter-free and is able to handle heterogeneous features that may not even be independent of each other. Finally, it introduces the notion of dyscrasia that allows us to detect not only which features are important in a particular task but also how they interact with each other. • Performing model-agnostic Explainable AI on classifiers. • Introducing the concept of dyscrasia of feature occurrences. • Assessing the relevance of a feature during classification. [ABSTRACT FROM AUTHOR]

Automatic modulation recognition (AMR) has garnered significant attention in both civilian and military domains, with applications ranging from spectrum sensing and cognitive radio (CR) to the deterrence of adversary communication. Index modulation (IM) represents an innovative digital modulation technique that exploits the indices of parameters of communication systems to transmit extra information bits. This paper aims to examine the performance of a convolutional neural network (CNN)-based AMR across various IM systems, including spatial modulation (SM), quadrature spatial modulation (QSM), and generalized spatial modulation (GSM) with eight digital modulation schemes. In this study, we leverage confusion matrices, receiver operating characteristic (ROC) curves, and F1 scores to illustrate the recognition model's outputs. • Modulation recognition models exploit deep learning algorithms effectively. • Index modulation is a novel technique for conveying additional information. • Three datasets are created for the recognition of eight digital modulation schemes. • Three modulation recognition scenarios analyze the system's performance. • Quadrature amplitude modulation types are more robust than phase shift keying types. [ABSTRACT FROM AUTHOR]

Financial fraud is a persistent problem in the finance industry that may have severe consequences for individuals, businesses, and economies. Graph Neural Networks (GNNs) are a class of deep learning models designed to operate on graph data structures that consist of nodes and edges connecting them. GNNs have emerged as a powerful tool for detecting fraudulent activities in complex financial systems because they can analyze the network structure of financial transactions, capturing patterns and anomalies that traditional rule-based and machine learning methods might miss. The objective of this systematic review is to provide a comprehensive overview of the current state-of-the-art technologies in using Graph Neural Networks (GNNs) for financial fraud detection, identify the gaps and limitations in the existing research, and suggest potential directions for future research. We searched five academic databases, including Web of Science, Scopus, IEEE Xplore, ACM, and science direct using specific keywords and search strings related to graph neural networks, financial areas, and anomaly detection to identify relevant publications, resulting in a total of 388 unique articles. We selected the relevant publications based on the inclusion, exclusion, and quality assessment criteria, and 33 articles were included in the review. In addition, forward snowballing was used to identify relevant papers that were not captured in the initial search. Data was extracted from the selected articles, then analyzed and summarized to identify current state, gaps, and trends in the literature. Our review presents a new taxonomy of GNNs applied in financial fraud detection and identifies potential research directions in this field. We find that GNNs applied to financial fraud detection have mostly been employed in a supervised or semi-supervised manner, with limited exploration of unsupervised approaches. In addition to financial areas, we explore the different types of graphs such as homogeneous, heterogenous, static, temporal, and dynamic graphs, and investigate the various learning mechanisms and anomaly types studied. We also note a lack of research on edge-level and graph-level anomaly detection commonly employed in financial domain. [ABSTRACT FROM AUTHOR]

Researchers and practitioners have attempted to predict the financial market by analyzing textual (e.g., news articles and social media) and numeric data (e.g., hourly stock prices, and moving averages). Among textual data, while many papers have been published that analyze social media, news content has gained limited attention in predicting the stock market. Acknowledging that news is critical in predicting the stock market, the focus of this systematic review is on papers investigating machine learning and text mining techniques to predict the stock market using news. Using Kitchenham's methodology, we present a systematic review of the literature on intelligent financial market prediction, examining data mining and machine learning approaches and the employed datasets. From five digital libraries, we identified 61 studies from 2015–2022 for synthesis and interpretation. We present notable gaps and barriers to predicting financial markets, then recommend future research scopes. Various input data, including numerical (stock prices and technical indicators) and textual data (news text and sentiment), have been employed for news-based stock market prediction. News data collection can be costly and time-consuming: most studies have used custom crawlers to gather news articles; however, there are financial news databases available that could significantly facilitate news collection. Furthermore, although most datasets have covered fewer than 100K records, deep learning and more sophisticated artificial neural networks can process enormous datasets faster, improving future model performance. There is a growing trend toward using artificial neural networks, particularly recurrent neural networks and deep learning models, from 2018 to 2021. Furthermore, regression and gradient-boosting models have been developed for stock market prediction during the last four years. Although word embedding approaches for feature representation have been employed recently with good accuracy, emerging language models may be a focus for future research. Advanced natural language processing methods like transformers have undeniably contributed to intelligent stock market prediction. However, stock market prediction has not yet taken full advantage of them. • Emerging articles (2015–2021) on news-based stock market prediction are reviewed. • There is a growing trend toward using news headlines for stock market prediction. • Recurrent neural networks are advantageous models for stock market prediction. • Stock market prediction not fully took advantage of Natural language processing yet. • Quarterly and annual reports of companies are authorized credible data sources. [ABSTRACT FROM AUTHOR]

Deep learning techniques for Sentiment Analysis have become very popular. They provide automatic feature extraction and both richer representation capabilities and better performance than traditional feature based techniques (i.e., surface methods). Traditional surface approaches are based on complex manually extracted features, and this extraction process is a fundamental question in feature driven methods. These long-established approaches can yield strong baselines, and their predictive capabilities can be used in conjunction with the arising deep learning methods. In this paper we seek to improve the performance of deep learning techniques integrating them with traditional surface approaches based on manually extracted features. The contributions of this paper are sixfold. First, we develop a deep learning based sentiment classifier using a word embeddings model and a linear machine learning algorithm. This classifier serves as a baseline to compare to subsequent results. Second, we propose two ensemble techniques which aggregate our baseline classifier with other surface classifiers widely used in Sentiment Analysis. Third, we also propose two models for combining both surface and deep features to merge information from several sources. Fourth, we introduce a taxonomy for classifying the different models found in the literature, as well as the ones we propose. Fifth, we conduct several experiments to compare the performance of these models with the deep learning baseline. For this, we use seven public datasets that were extracted from the microblogging and movie reviews domain. Finally, as a result, a statistical study confirms that the performance of these proposed models surpasses that of our original baseline on F1-Score. [ABSTRACT FROM AUTHOR]

Over the past decades, online social networks such as Twitter and Facebook have become a significant part of people's daily lives, particularly amid the ongoing global calamity — the COVID-19 pandemic. This gives room for social bot attacks that are designed to automatically replicate the behavior of real accounts. Most of these bots are employed for nefarious purposes such as disseminating false information, artificially amplifying the popularity of a person or movement, or spreading spam. Many studies have been conducted in an attempt to discover new strategies for identifying social bot accounts. To deal with large-scale attacks from social bots, Machine Learning has emerged as a noticeable path of bot detection problem that can handle massive amounts of data. However, the heterogeneity between studies in terms of problem statements, proposed processes, datasets, and evaluation metrics makes it difficult to assess and compare the efficiency of proposed methods. In this paper, we propose a systematic view of supervised learning methodologies for tweet-based social bot detection, ranging from shallow learning to specific deep learning models. In addition, we introduce a framework that measures various performance aspects and summarizes the in-depth analysis of the results, which were obtained using two datasets comprising 26224 labeled Twitter accounts. The results of this framework, we believe, will be beneficial as a practical guideline for other bot detection research or applications that require the use of machine learning techniques. [ABSTRACT FROM AUTHOR]

Nowadays, video surveillance systems are commonly found in most public and private spaces. These systems typically consist of a network of cameras that feed into a central node. However, the processing aspect is evolving towards distributed approaches, leveraging edge-computing. These distributed systems are capable of effectively addressing the detection of people or events at each individual node. Most of these systems, rely on the use of deep-learning and segmentation algorithms which enable them to achieve high performance, but usually with a significant computational cost, hindering real-time execution. This paper presents an approach for people detection and action recognition in the wild, optimized for running on the edge, and that is able to work in real-time, in an embedded platform. Human Action Recognition (HAR) is performed by using a Recurrent Neural Network (RNN), specifically a Long Short-Term Memory (LSTM). The input to the LSTM is an ad-hoc, lightweight feature vector obtained from the bounding box of each detected person in the video surveillance image. The resulting system is highly portable and easily scalable, providing a powerful tool for real-world video surveillance applications (in the wild and real-time action recognition). The proposal has been exhaustively evaluated and compared against other state-of-the-art (SOTA) proposals in five datasets, including four widely used (KTH, WEIZMAN, WVU, IXMAX) and a novel one (GBA) recorded in the wild, that includes several people performing different actions simultaneously. The obtained results validate the proposal, since it achieves SOTA accuracy within a much more complicated video surveillance real scenario, and using a lightweight embedded hardware. • Real-time action recognition based on LSTMs with an ultralight feature vector. • Experimental results on for widely used dataset and a novel dataset in-the-wild. • Comparison to other state-of-the-art methods. • Achieves accuracy close to state-of-the-art while reducing computational cost. • Proposal runs in real time in an embedded platform. [ABSTRACT FROM AUTHOR]

Accurate traffic congestion estimation and prediction are critical building blocks for smart trip planning and rerouting decisions in transportation systems. Over the decades, there have been many studies focusing on traffic congestion estimation and prediction with different statistical approaches (e.g., Markov chain) and machine learning models (e.g., clustering, Bayesian networks, and artificial neural networks). However, there is a lack of a unified framework to address the mechanisms of different models and integrate the advantages of different methods through combinations. This paper introduces the FD-Markov-LSTM model, a hybrid interpretable approach that combines the fundamental diagram (FD), Markov chain, and long short-term memory (LSTM). The aim is to estimate and predict traffic states by integrating statistical data in both congested and uncongested scenarios. The FD-Markov-LSTM model leverages the FD to identify hierarchical traffic states and utilizes the Markov process to capture the probabilistic transitions between these states. We employ the LSTM model to further capture the residual time series produced by the Markov chain model (assuming a memoryless property) to enhance the estimation and prediction performance. The proposed model's accuracy in estimating and predicting traffic flow is evaluated using empirical data from three case studies conducted in Beijing and Los Angeles. The results highlight a significant improvement in accuracy compared to classical benchmark models such as the Markov model, ARIMA model, k-Nearest Neighbor model, Random Forest model, and LSTM. Specifically, the FD-Markov-LSTM model achieves reductions of over 39% in mean absolute error, 35% in root mean squared error, and 7.4% in mean absolute percentage error. These results clearly demonstrate that the FD-Markov-LSTM model outperforms the benchmark models, enabling more precise predictions of traffic flow. [ABSTRACT FROM AUTHOR]

Simulation-driven structural health monitoring (SHM) is crucial for enhancing the safety and reliability of structures. Its primary aim is to utilize numerical simulations to design efficient SHM systems, reducing costs while maintaining effectiveness. However, challenges arise due to disparities between simulations and physical measurements, stemming from uncertainties, noise, and simplifications. These challenges also extend to SHM methods using machine learning, particularly deep neural networks (DNNs). This paper bridges the gap between DNNs trained on simulated data and real-world experiments by investigating three transfer learning (TL) techniques: feature augmentation, direct fine-tuning, and boosting-based fine-tuning. The focus is on guided wave damage interaction coefficients (WDICs) as damage-sensitive features, modeled by DNNs for damage identification in thin plates, commonly found in aerospace structures like aircraft fuselages. This classification task compares measured and predicted WDICs in a common guided wave SHM configuration with one actuator and three sensors. Training and extensive testing datasets are generated using advanced finite element simulations and laser Doppler vibrometer experiments. The study demonstrates that TL techniques significantly enhance DNN prediction accuracy and improve damage identification performance for untrained scenarios compared to purely simulation-based DNNs. Among the TL approaches explored, feature augmentation with appropriate activation functions and boosting-based fine-tuning yield the most promising results. Using four out of twelve TL datasets, feature augmentation reduces the simulation-based mean absolute percentage error from 46.80% to 35.98% and increases coefficient of determination R 2 values from 0.02 to 0.72. With six TL datasets and a boosting-based fine tuning, the mean absolute percentage error decreases from 47.42% to 35.20%, maintaining similar R 2 values. While TL substantially enhances prediction accuracy, its impact on damage identification is slightly less pronounced, with improvements in relative classification accuracy by approximately 5% compared to simulation-based DNNs. These findings highlight the efficacy of TL techniques for damage identification in thin plates using WDICs and DNNs, emphasizing the influence of TL training and DNN model variations. Furthermore, they showcase the potential of TL methods to enhance DNN performance in future SHM applications, particularly emphasizing the promise of feature augmentation and boosting. • Novel simulation-driven structural health monitoring technique using guided waves. • Deep neural network modeling to predict wave damage interaction coefficients. • Bridging disparity between simulations and experiments through transfer learning. • Study of transfer learning techniques based on feature augmentation and fine tuning. • Demonstrated significant enhancement in simulation-based neural network performance. [ABSTRACT FROM AUTHOR]

At present, the serious challenge of human-robot collaboration is how to establish an interaction interface with multi-directional interactivity and robustness. This paper applies surface electromyography (sEMG) signals as the interface medium, and proposes a robust 3D arm strength estimation model based on hybrid deep learning networks (R3DNet) to build a stable interaction interface and achieve reliable human-robot collaboration performance in 3D operation scenarios. R3DNet adopts the modular design idea and includes a feature extraction module, a regression estimation module, and a robust enhancement module. In the feature extraction module, an autoencoder structure-based network and a new loss function are proposed to extract the fusion features of multi-channel sEMG signals. In the regression estimation module, an arm strength model based on transfer learning algorithms is proposed. This model considers the similarity of arm strength estimation in each direction and realizes the arm strength estimation from one to three dimensions through the fine-tune strategy. The robust enhancement module adopts a residual structure and aims to address the problem of poor model robustness due to non-ideal conditions such as electrode displacement, muscle fatigue, muscle time-varying characteristics and differences in muscle characteristics among subjects. Further, an improved adaptive admittance control method is proposed to help the robot to adjust collaborative behavior according to the changing arm strength of the human tutor. The results of comparison experiments and multi-point flexible assembly experiments verify the validity and engineering realizability of R3DNet for human-robot collaboration. [ABSTRACT FROM AUTHOR]

• AER system utilizes kinematic data for enhanced human–computer interaction. • Diverse algorithms achieve >99% success (small windows), >88% (larger). • Innovative analysis: joint neighborhood, window size, axis combination efficiency. • Practical CNN algorithms enable real-time emotion recognition. • Insights from transparent datasets lead to emotion recognition advancements. This study introduces an automatic emotion recognition system (AER) focusing on skeletal-based kinematic datasets for enhanced human–computer interaction. Departing from conventional approaches, it achieves real-time emotion recognition in real-life situations. The dataset covers seven emotions and undergoes assessment by eight diverse machine and deep learning algorithms. A thorough investigation is undertaken by varying window sizes and data states, including raw positions and feature-extracted data. The findings imply that incorporating advanced techniques like joint-related feature extraction and robust classifier models yields promising outcomes. Dataset augmentation via varying window sizes enriches insights into real-world scenarios. Evaluations exhibit classification accuracy surpassing 99% for small windows, 94% for medium, and exceeding 88% for larger windows, thereby confirming the robust nature of the approach. Furthermore, we highlight window size's impact on emotion detection and the benefits of combining coordinate axes for efficiency and accuracy. The analysis intricately examines the contributions of features at both the joint and axis levels, assisting in making well-informed selections. The study's contributions include carefully curated datasets, transparent code, and models, all of which ensure the possibility of replication. The paper establishes a benchmark that bridges theory and practicality, solidifying the proposed approach's effectiveness in balancing accuracy and efficiency. By pioneering advanced AER through kinematic data, it sets a new standard for efficacy while driving seamless human–computer interaction through rigorous analysis and strategic design. [ABSTRACT FROM AUTHOR]

Deep machine learning models including Convolutional Neural Networks (CNN) have been successful in the detection of Mild Cognitive Impairment (MCI) using medical images, questionnaires, and videos. This paper proposes a novel Multi-branch Classifier-Video Vision Transformer (MC-ViViT) model to distinguish MCI from those with normal cognition by analyzing facial features. The data comes from the I-CONECT, a behavioral intervention trial aimed at improving cognitive function by providing frequent video chats. MC-ViViT extracts spatiotemporal features of videos in one branch and augments representations by the MC module. The I-CONECT dataset is challenging as the dataset is imbalanced containing Hard-Easy and Positive–Negative samples, which impedes the performance of MC-ViViT. We propose a loss function for Hard-Easy and Positive–Negative Samples (HP Loss) by combining Focal loss and AD-CORRE loss to address the imbalanced problem. Our experimental results on the I-CONECT dataset show the great potential of MC-ViViT in predicting MCI with a high accuracy of 90.63% accuracy on some of the interview videos. • We propose MC-ViViT to detect MCI from facial videos in the I-CONECT Study. • We design the MC module to enrich the extracted spatio-temporal features. • We develop HP loss to address the inter-and intra-class imbalanced data. [ABSTRACT FROM AUTHOR]

Automated Essay Scoring (AES) is a challenging topic in Natural Language Processing. Recently, deep learning models have achieved remarkable performance for the AES task. However, applying deep learning models to the AES system in practice is expensive when both data collection and model training are taken into consideration. This paper aims to tackle this problem by proposing the Crowdsourcing-based Automated Essay Scoring (CAES) framework. The proposed framework gradually collects data through crowdsourcing and incrementally trains the AES models. In particular, we propose the Incremental Learning with Dynamic Exemplar Herding (ILDEH) approach to simultaneously tackle catastrophic forgetting and concept drift. The proposed approach dynamically updates the exemplar set by the Dynamic Exemplar Herding algorithm to obtain the best approximation of the overall data distribution and selectively apply knowledge distillation on the model outputs by Linear Outlier Suppression loss to retain the learned knowledge. Moreover, we use a lightweight AES model for effective and efficient essay scoring. The experimental results show that our proposed ILDEH approach outperforms other strong baseline approaches for the AES task. Moreover, the CAES framework is able to steadily improve the AES performance in the crowdsourcing environment with only 10.6% training time of the conventional approach. Further analysis shows that one single CPU server can support daily updates of more than 300 AES models, which is sufficient for most practical AES systems. • We proposed the Crowdsourcing-based Automated Essay Scoring (CAES) framework. • We proposed an incremental learning approach to efficiently train AES models. • Our proposed approach outperforms strong baseline approaches. • The CAES framework requires only 10.6% of the conventional training time. • The CAES framework supports daily update of 300 AES models on a single CPU server. [ABSTRACT FROM AUTHOR]

Animation drama and movies are important entertainment sources for children and young people. However, many of these entertainment videos introduce ideas such as violence, fighting, abuse, and car accidents, which are not suitable for children and young people. In order to prevent children from viewing these videos, different countries' government bodies put restrictions on age on these videos. Violence detection is mostly done using manual inspection, which is time and resource consuming and cannot achieve a very good result. Therefore, artificial technologies such as deep learning and machine learning can detect violence in a video. Bleeding is one of the distinct characteristics of violent actions presented in such videos. This paper aims to propose a system that can detect a given video's blood in real-time. Researchers have proposed different image processing-based techniques to detect violence, but their proposed systems lack precision and real-time detection. To this end, a deep learning-based approach is proposed to detect violence in a given video and images. A Faster R-CNN model is modified due to the sophisticated characteristics of the violence in the cartoon and animation images to achieve the highest performance. The backbone of the proposed violence detection model is changed to a modified RegNet model to extract the features from the frame. The standard inner lateral connection is replaced with the modulated deformable convolutional (MDC) layer to extract deformable feature maps. A novel distributed attention module (DAM) is proposed on the feature pyramid network to improve the performance of the feature extraction. The multiscale Region of interest (ROI) Align is adapted to improve the detection performance of violence in a different scenario. Moreover, the classification approach is integrated with the detection model to classify the different levels of violence in a given frame. We have explored state-of-the-art object detection methods such as Faster R-CNN, Cascade R-CNN, yolov3-spp, SSD, FCOS, and yolov5 to detect violence in a particular frame of a video. We compared the different models' performances and found out that the modified Faster R-CNN is appropriate for real-time blood detection in a given video or image with high accuracy. This study makes a significant contribution to detecting violence in animation videos which will help different platforms and government bodies to regulate entertainment content. [ABSTRACT FROM AUTHOR]

The human skin, the largest organ with multiple layered functionalities, houses melanocytes in the deeper strata of its epidermis. These cells can be adversely impacted by ultraviolet radiation, thereby instigating melanoma, the deadliest form of skin cancer. Failure to detect melanoma at an early stage can potentially lead to metastasis, forming complex tumors in other tissues. Despite substantial efforts, visual inspections can occasionally overlook melanoma cases due to inherent subjectivity. To surmount this challenge, an automated detection system is necessary. Recent attempts to establish such a system have predominantly employed "push-through" strategies involving deep (neural) networks and their ensembles, which however necessitate significant computational resources. This paper presents a novel approach, amalgamating a conventional machine learning technique, Bag of Visual Words, with a pretrained deep neural network for comprehensive deep feature extraction from enhanced input image patches. The proposed method, assessed on the ISIC Challenge 2017 dataset, surpassed all other entries on the challenge leader-board, registering an accuracy of 96.2% in the task of lesion classification. • Novel method for melanoma detection combining Bag of Visual Words with a deep network. • Patch-based feature extraction from lesion images integrated into BoVW with K-means. • Successful image enhancement technique for feature extraction from lesion images. • Efficient framework: modular architecture, easy updates, less training delay. [ABSTRACT FROM AUTHOR]

A metaheuristic-based structural optimization method, whilst being popularly adopted due to its advantages in by-passing gradient function calculations, requires the use of time-consuming advanced analyses for constraint evaluation. To overcome this drawback, the present paper proposes a robust (machine learning-based) optimization method that combines the light gradient boosting machine (LightGBM) with the efficient p-best differential evolution (E p DE) method. In essence, the LightGBM classification model is constructed to assess the constraint (safety and integrity) satisfaction of structures. An efficient framework using a so-called safety parameter is proposed to prevent inaccurate predictions of the LightGBM model. The E p DE processes the optimization procedures on the constructed classification LightGBM model. This enables an enhanced machine learning-based optimization technique that not only maintains the sufficiently accurate optimal design of structures but also significantly reduces the required computing efforts, as compared to standard optimization schemes. Various examples of steel structure designs (i.e., two of which have been provided herein) have been successfully performed by the proposed approach. These illustrate the accuracy and robustness of the proposed method, where good comparisons with reference algorithms (including standard DE with "DE/rand/1" mutational strategy, Jaya, Rao-1 and CaDE) are evidenced. The statistical values collected present the accuracy and reliability of the proposed method in obtaining the minimum total weight designs with substantial (some 60%) reduction of computing efforts required to furnish optimal results. [ABSTRACT FROM AUTHOR]

Deep learning (DL) has made its way into many disciplines ranging from health care to self-driving cars. In financial markets, we see a rich literature for DL applications. Particularly, investors require robust algorithms that can navigate and make sense of extremely noisy and volatile markets. In this work, we use deep learning to select a portfolio of stocks and use a genetic algorithm to optimize the hyperparameters of DL. The work analyzes the improvement in using genetic-based hyperparameter optimization over grid searches. The Genetic Algorithm brings 40% improvements in prediction when compared to a random-grid search. Novelty-wise, the work couples a genetic-based hyperparameter optimization with multiple Deep RankNet models to predict the behavior of financial assets. Our results show promising portfolio returns 20% better than the general market. In the highly volatile COVID 19 period, the models exceed market returns by more than double. Overall, this paper brings a comprehensive work that integrates hyperparameter optimization, Deep RankNet, LSTM, period size variations, input variable transformation, feature selection, training/evaluation ratio analysis, and multiple portfolio selection strategies. [ABSTRACT FROM AUTHOR]

The Computational Fluid Dynamics (CFD) simulations at different fidelity levels are a common tool for predicting the aerodynamic performance in many engineering applications such as; automotive and aerospace to cite a few. However, the associated computational costs with these simulations are not trivial, and using the CFD simulations might easily become unpractical specially at the early design stages. The number of the required CFD simulations could be reduced immensely by using Artificial Intelligence (AI) methods which can predict, for example, the aerodynamic coefficients based on previous and limited CFD simulations. Having discussed the AI potential to minimize the CFD usage, the objective of this work is to investigate and compare the ability of different AI methods to predict the aerodynamic coefficients using the same CFD database from the literature. In this study, three different AI methods, namely; machine learning (e.g. Extreme Gradient Boosting (XGB), CatBoost, Bagging, Light Gradient Boosting Machine (Light-GBM), Random Forest (RFR), Gradient Boosting), deep learning (e.g. One Dimensional Convolutional Neural Network (Conv1D)) and surrogate models (e.g. Gaussian Process Regression (GPR)) have been utilized to estimate a sample CFD database (e.g. NACA0012, RAE2822) from the existing literature, and the prediction performance of these methods have been compared. The results show that, the aerodynamic coefficients predicted using CatBoost, XGB and Bagging are in good agreement with the reference CFD results. Furthermore, the proposed Conv1D method with LeakyReLU activation function showed a promising results in the development of aerodynamic models which can be used in the early stages of the air vehicles design. The obtained results from this work indicate that the CatBoost and the XGB models required much smaller training time compared to the Conv1D method, moreover, the proposed regression trees; CatBoost, Bagging and XGB can reduce the required number of CFD simulations significantly. • Various artificial intelligence methods were studied to predict aerodynamic data. • Regression tree models such as XGB, Catboost and LightGBM were used. • Surrogate models and deep learning models (i.e. Conv1D) also were used. • Detailed comparison were presented between the aforementioned models. • The paper suggests the suitability of XGB and Catboost for aerodynamics cases. [ABSTRACT FROM AUTHOR]

The past decade has witnessed the rapid development of Artificial Intelligence (AI), especially the explosion of deep learning-related connectionist approaches. This study combines traditional literature review, bibliometric methods, and the Science of Science (SciSci) theory to scrutinize the development context of AI in the last decade on AMiner. 4 4 www.aminer.cn , AMiner is an academic mining system, which will be introduced in the following section. With the assistance of AMiner tools and datasets, this paper aims to describe a further explicit context and evolution of AI in the past decade from the development of connectionist approaches. Five aspects of the past decade are highlighted: self-learning and self-coding algorithms, Recurrent Neural Networks (RNN) algorithms, reinforcement learning, pre-trained models, and other typical deep learning algorithms, which represent the significant progress of this field. By combining these critical parts, we then summarize the current limitations and corresponding future of AI trends in the next decade and discuss some topics about the next generation of AI. Discoveries in this paper will benefit AI research in promoting understanding of the current critical stage and future trends of AI development and the AI industry in the dramatic ascendant for the academic research results transformation and its industrial layout. [ABSTRACT FROM AUTHOR]

Globally, all the major crops are significantly affected by diseases every year, as manual inspection across diverse fields is time-consuming, tedious, and requires expert knowledge. This leads to significant crop loss in different parts of the world. To provide effective solutions, several smart agriculture solutions are deployed for the control of pests and plant diseases using vision-based machine learning techniques. Despite rapid growth in the field, not many methods have been explored for their suitability in real-time applications. Several open challenges need to be addressed for the applicability of machine learning techniques in IoT-based smart agriculture solutions. Starting from data capturing methods and the availability of public datasets, the present paper provides a comprehensive review of vision-based machine learning techniques for plant disease detection. Initially, 1337 articles were selected from various scholarly resources to perform the survey. Based on the saliency of approaches, 148 articles are reviewed in this paper. Interestingly, a significant amount of research in this direction is taken up by Chinese and Indian researchers, and deep learning is the current research trend, as in other fields. The review concludes that a majority of existing methods exhibit their efficacy on public datasets captured mostly in controlled environmental conditions, but their generalization capability for in-field plant disease detection has not been explored. Lightweight CNN-based methods, on the other hand, have been designed for a limited number of diseases only, and are generally trained on small datasets. The scarcity of large-scale, in-field public datasets is one of the major bottlenecks in developing solutions that can work for a wide variety of plant diseases. [ABSTRACT FROM AUTHOR]

Vertical federated learning (VFL) is a privacy preserving collaborative machine learning technique designed for distributed learning scenarios in which data from different parties have overlap in the sample space. In this paper, a VFL method for feature selection, which is an effective dimensionality reduction technique that selects a subset of informative features from high-dimensional data by eliminating irrelevant and redundant features, is proposed. Because of the potential insufficiency of useful information for learning informative features and the difficulty in sharing raw data among parties due to the increasing awareness of data privacy protection, it is desirable to exploit information from multiple parties without raw data sharing. In this paper, we propose a VFL-based feature selection method that leverages deep learning models as well as complementary information from features in the same samples at multiple parties without data disclosure. In order to further improve feature selection performance, information of samples that do not have features appearing in all parties are also utilized. Promising results in extensive experiments show the effectiveness of the proposed approach in terms of collaborative feature selection without data sharing. • In this paper, we bridge this gap by proposing a novel VFL-based feature selection method—Vertical Federated Learning-based Feature Selection (VFLFS). To the best of our knowledge, this is the first deep learning-based vertical federated learning approach with feature selection. • A strategy to make use of non-overlapping samples is also proposed to improve feature selection effectiveness. • The proposed VFLFS approach has been evaluated extensively based on real-world datasets. The results show that VFLFS can significantly improve model performance under VFL settings compared to four state of the art baselines, especially in conditions where a large proportion of the data samples do not overlap across data owners. [ABSTRACT FROM AUTHOR]

Drug design is a research process with a goal of creating a chemical drug to produce the desired biological effect. Because of the long time and the high cost issues associated with traditional drug discovery, there is a need to develop new techniques and strategies to increase the diminishing effectiveness of traditional approaches. Ligand-Based Virtual Screening (LBVS) plays a vital role in the early stage of the drug discovery. It could constitute a possible solution to solve the time and cost problems. Subsequently, the researchers are looking for new methods to find new active compounds and bring them to market in a short time. LBVS can be enhanced by different methods and strategies such as Machine Learning and Deep Learning. In this paper, a Deep Convolutional Neural Network method is proposed to enhance the performances of Ligand-Based Virtual Screening process (DCNNLB). Two main contributions are presented in this paper, The first contribution consists of designing a model based on Deep Convolutional Neural Network (DCNN) for LBVS. We propose several topological network models to find the one that gives the best performance such as accuracy and recall. For this, many network topology configurations have been proposed, and a variety of parameters have been taken into account. Furthermore, our proposed model is trained on all compounds of all activity classes of the Drug Data Report Database (MDDR). Thus, it presented a mean accuracy of 0.98 for all three MDDR Datasets. The second contribution is to generate a new learning representation in order to better represent chemical compound. This representation is based on the extraction of the automatic features learning from the weights of our proposed model. Consequently, it is very efficient in calculating molecular similarity and performances of the LBVS process. The obtained results with the three different datasets drawn from the MDDR and the performance evaluation with ANOVA test, have proved the superiority in performance of our proposed method compared to the different conventional methods. [Display omitted] • New molecular representations obtained by the proposed deep convolutional model. • Enhancement in effectiveness of similarity searching with DCNNLB method in LBVS. • Decreasing in training time by using the molecular learning representation. • The proposed DCNNLB method significantly outperforms the conventional TAN method. [ABSTRACT FROM AUTHOR]

• A embedding method MAHE-IM for IM on heterogeneous network. • MAHE-IM extracts and aggregates multiple heterogeneous relationships. • Eight network embedding and seven GNN methods extended for IM. • MAHE-IM outperforms eighteen state-of-the-art algorithms. • A web server (https://mahe-im.com/) is developed for maximizing user convenience. Influence maximization (IM), as an essential problem in social network analysis, can identify a minimum group of the influential nodes to maximize the spread of information on the network. The majority of IM studies focus on homogeneous networks, whereas in real world, heterogeneous networks are ubiquitous. The existing IM methods based on homogeneous networks do not consider a variety of complex heterogeneous relationships and the attribute of different types of nodes, which don't accommodate most real scenario. It is of great significance to study IM methods based on heterogeneous networks, where efficient integrating complex multiple semantic relationships and structures embedded in the heterogeneous network is the key breakthrough point. In this paper, a novel deep learning algorithm for influence maximization on heterogeneous networks based on a multiple aggregation of heterogeneous relation embedding named MAHE-IM is proposed, which can capture the heterogeneous high-order structure and semantic features of heterogeneous information networks. For more comprehensive and systematical evaluation of MAHE-IM, we further extend fourteen state-of-the-art homogeneous and heterogeneous network embedding and graph neural network methods for IM problem and propose fourteen extented IM algorithms. Four popular IM algorithms and our extended fourteen IM algorithms are taken as eighteen baseline algorithms, which can be categorized three types: greedy-based, network embedding-based and GNN-based IM algorithms. Compared with eighteen baseline algorithms, the experimental results illustrate that MAHE-IM is significantly efficient and effective. MAHE-IM is more practical on heterogeneous networks. In addition, in order to maximize the convenience for users, a webserver (https://mahe-im.com/) is developed. The users can obtain the IM results by submitting their heterogeneous networks to the webserver. The corresponding source code and used data in this paper can also be available at https://mahe-im.com/ and https://codeocean.com/capsule/4091031/tree/v1. [ABSTRACT FROM AUTHOR]

• Higher accuracy of Alzheimer diagnosis system improves earlier diagnosis chances. • Physical activity offers substantial resistance to progression of Alzheimer's disease. • Ability to classify and log activities can help physicians and Alzheimer's patients. • Machine based activity monitoring offers improved Alzheimer's patient privacy. Machine based analysis and prediction systems are widely used for diagnosis of Alzheimer's Disease (AD). However, lower accuracy of existing techniques and lack of post diagnosis monitoring systems limit the scope of such studies. In this paper, a novel machine learning based diagnosis and monitoring of AD-like diseases is proposed. The AD-like diseases diagnosis process is accomplished by analysing the magnetic resonance imaging (MRI) scans using deep learning and is followed by an activity monitoring framework to monitor the subjects' activities of daily living using body worn inertial sensors. The activity monitoring provides an assistive framework in daily life activities and evaluates vulnerability of the patients based on the activity level. The AD diagnosis results show up to 82% improvement in comparison to well-known existing techniques. Moreover, above 95% accuracy is achieved to classify the activities of daily living which is quite encouraging in terms of monitoring the activity profile of the subject. [ABSTRACT FROM AUTHOR]

Visibility in foggy weather is of great value for traffic management and pollution monitoring. However, vision-based fog visibility estimation methods are usually based on a single image to approximate the visibility in foggy weather, and most existing data-driven machine learning models struggle to capture effective features and achieve high estimation accuracy due to the severe image degradation caused by reduced visibility and lack of real scene images. Therefore, this paper proposes a novel deep learning framework based on multi visual feature fusion for fog visibility estimation, named VENet, which comprises of two subtask networks (for fog level classification and fog visibility estimation) constructed in a cascade structure. A special feature extractor and an anchor-based regression method (ARM) are proposed to help improve the accuracy. Further, a standard Fog Visibility Estimation Image (FVEI) dataset containing 15,000 images of real fog scenes is established. This dataset greatly bridges the lack of suitable data in the field of vision-based visibility estimation. Extensive experiments have been conducted to demonstrate the performance of the proposed VENet, where the error of fog visibility estimation is less than 5% at 500 m and the fog level classification accuracy is at least 92.3%. In addition, the proposed VENet has been applied on Yunnan Xiangli and Mazhao Expressway surveillance with promising performance in practice. [ABSTRACT FROM AUTHOR]

• A class incremental learning framework is proposed for open sets. • The framework is implemented based on multiple one-class classifiers (OCCs). • An improved deep SVDD is proposed as the basic OCC of the framework. • Several tricks (e.g., feature fusion, band attention) are used to improve OCC. • HSI combined with this method has broad application prospects in seed industry. Rapid and non-destructive classification of seed varieties is one of the important goals pursued by modern seed industry. Due to the large number of maize varieties in practice, collecting training samples that exhaust all varieties to train a classifier is extremely difficult. Therefore, maize seed classification in the real world faces the challenge of variety renewal and rejection of unknown varieties. This paper proposes an end-to-end trainable incremental learning (IL) framework based on hypercube data. This method achieves class-IL via learning one-class classifier (OCC) incrementally, and directly uses raw data as input without additional data preprocessing and feature extraction. The OCC is a dual deep support vector data description, which makes full use of spectral and spatial information to establish an exclusive hypersphere for a specific variety to receive the variety and reject unknown varieties. To remove the interference of redundant bands, a band attention and sparse constraint module is added to automatically assign the weights of redundant bands to zero, thereby maximally improving the performance of the model. Moreover, a new loss function is defined to alleviate the difficulty of parameter updating after sparse constraint. Experimental results on our open set indicate that the accuracy of the proposed method for receiving known varieties and rejecting unknown varieties are both above 91 %, which has significant advantages over the other two state-of-the-art IL methods. In the future, the corresponding OCCs can also be deleted from the whole framework according to the varieties eliminated by the government to reduce computational overhead and inter-class interference. Overall, the proposed method can perform both IL and open set recognition. [ABSTRACT FROM AUTHOR]

The continuous malicious software (malware) attacks on smartphones pose a serious threat to the security of users, especially the dominant platform Android. Data-driven methods based on machine learning algorithms are a promising way to defend against that. In this paper, we explore the limitations of this kind of methods in improving the performance of malware detection, e.g., considering each feature in isolation and relying on balanced class data, and propose a multi-model ensemble framework MEFDroid by combining individual predictors, where hybrid deep learning based feature extraction methods are adopted to learn meaningful features from raw data. Besides, a novel fusion scheme is exploited to fuse multi-level representations and mine their correlations to maximize the utilization of original information. To evaluate the effectiveness of the MEFDroid framework, we conduct a step-by-step model justification experiment to investigate how the proposed algorithms (i.e., ESAES, EDAES and EDAFS) enhance the malware detection performance gradually. We also compare the proposed algorithms with classical machine learning methods, conventional sampling solutions for the imbalance problem. Besides, we investigate the reliability of our proposed methods using another public dataset. Our extensive experimental results demonstrate that the target model EDAFS in MEFDroid outperforms others in terms of most metrics, which means that it is an effective alternative solution to detect Android malware. [ABSTRACT FROM AUTHOR]