To address the challenge of autonomous navigation for mobile robots in complex environments, many recent path planning methods employ a two-layer motion framework that integrates both global and local planning. However, global planning algorithms based on evolutionary algorithms often yield suboptimal paths, primarily due to premature convergence and wasted iterations in infeasible regions. Meanwhile, local planning algorithms, particularly the dynamic window approach (DWA) and its variants, are frequently tailored to specific scenarios, limiting their generalizability. To address these issues, this paper proposes an effective path planning method, namely M2PP. The global planning algorithm in M2PP enhances particle swarm optimization (PSO) by introducing multiple new search phases, creating a multi-phase PSO that can quickly identify feasible global paths and thoroughly explore the sampling space. To effectively handle multiple scenarios in a more generalized manner, the local planning component, multi-scenario adaptative DWA, integrates two novel terms into its cost function, enhancing both static and dynamic obstacle avoidance. Extensive simulations demonstrate that multi-phase PSO exhibits strong performance and robustness, especially in highly complex scenarios, and multi-scenario adaptative DWA can generate safe and efficient local trajectories in hazardous conditions. Additionally, experiments deploying the M2PP method on a real-world mobile robot further confirm its feasibility.

Learning high-level representations for graphs is crucial for tasks like node classification, where graph pooling aggregates node features to provide a holistic view that enhances predictive performance. Despite numerous methods that have been proposed in this promising and rapidly developing research field, most efforts to generalize the pooling operation to graphs are primarily performance-driven, with fairness issues largely overlooked: i) the process of graph pooling could exacerbate disparities in distribution among various subgroups; ii) the resultant graph structure augmentation may inadvertently strengthen intra-group connectivity, leading to unintended inter-group isolation. To this end, this paper extends the initial effort on fair graph pooling to the development of fair graph neural networks, while also providing a unified framework to collectively address group and individual graph fairness. Our experimental evaluations on multiple datasets demonstrate that the proposed method not only outperforms state-of-the-art baselines in terms of fairness but also achieves comparable predictive performance.

While automatic anthropometric measurement extraction has witnessed growth in recent years, effective, non-contact, and precise measurement methods for dressed humans in arbitrary poses are still lacking, limiting the widespread application of this technology. The occlusion caused by clothing and the adverse influence of posture on body shape significantly increase the complexity of this task. Additionally, current methods often assume the availability of a complete 3D body mesh in a canonical pose (e.g., ”A” or ”T” pose), which is not always the case in practice. To address these challenges, we propose MeasureXpert, a novel learningbased model that requires only two unregistered, partial, and dressed body scans as input, and accommodates entirely independent and arbitrary poses for each scan. MeasureXpert computes a comprehensive representation of the naked body shape by synergistically fusing features from the front- and back-view partial point clouds. The comprehensive representation obtained is mapped onto a 3D undressed body shape space, assuming a canonical posture and incorporating predefined measurement landmarks. A pointbased offset optimization is also developed to refine the reconstructed complete body shape, enabling accurate regression of measurement values. To train the proposed model, a new large-scale dataset, consisting of 300K samples, was synthesized. The proposed model was validated using two publicly available real-world datasets and was compared with different relevant methods. Extensive experimental results demonstrate that MeasureXpert achieves superior performance compared to the reference methods. The code and dataset are available at: MeasureXpertProject.

Accurate assignment of International Classification of Diseases (ICD) codes is a challenging task due to a large label space, fine-grained distinctions between related codes, and the unstructured nature of clinical texts. While most prior work formulates this problem as multi-label classification at the document level, we explore a less-visited alternative perspective: multiclass ICD prediction at the evidence level. We constructed a silver-labeled dataset of evidence-code pairs from the MIMIC-IV corpus and trained a multiclass classifier based on Bio ClinicalBERT to predict a single ICD code for each evidence clause. Our experiments on the MDACE and CodiEsp datasets demonstrate high accuracy, with most errors involving clinically related or overlapping codes. The CodiEsp results highlight dataset challenges due to overcoding. Although not designed to compete with document-level systems, our approach produces interpretable and detailed predictions and can serve as a modular component in larger automated coding pipelines. We argue that evidence-level multiclass classifiers, especially with rare code augmentation, represent a valuable step toward transparent and hybrid clinical coding systems. The code is publicly available on GitHub1.

The emotion-cause pair extraction (ECPE) task is to simultaneously extract emotions and causes as pairs (EC-pairs) from documents, which is important for natural language processing. Previous research tackled this task via a two-step approach, which first predicts separately the emotion and cause clauses, and then pairs them up by using a binary classifier. However, such a two-step approach may suffer from the possible propagation of errors, and it neglects the interaction between emotions and causes. In this article, an end-to-end double-graph method with relational enhancement (DGRE) is proposed to stimulate two relationship modes among clauses, i.e., semantic dependence and logical dependence. First, two united graph encoders are established to embed the semantic dependence into the representation of clauses and pairs. The first encoder is built on graph attention networks (GATs) for clause-level representation, the result of which is used by a relational graph convolutional network (RGCN) for the refinement of pair-level representation. Aiming to enhance the fitting ability of logical dependence, the emotion-type classification task is introduced into the multitask learning framework of GATs, which can effectively distinguish the logical relations between clauses according to their emotion types. Moreover, seven types of dependence relations have been designed for the node connections in RGCN, which emphasize the contextual interaction and clustering among neighboring nodes. Experiments on a benchmark Chinese corpus demonstrate that the proposed DGRE approach could effectively establish the communication mechanism between clauses and pairs from multiple perspectives, and comparisons with state-of-the-art (SOTA) models well validate its effectiveness.

Developing near-accurate crop type mapping is crucial for effective agricultural monitoring. However, this is often challenged by the lack of labeled data in many regions and the issue of domain shift. To address these challenges, we propose a domain adaptation framework that leverages transformer-based time-frequency feature extraction to create domain-invariant representations, enabling reliable crop type mapping in unlabeled target regions. The framework incorporates a transformer encoder along with a frequency encoder to capture the temporal and frequency-domain pattern of the time series. We use three main loss functions, namely the Reconstruction loss, the Classification loss and the Alignment loss during the training of the network, which ensures the model simultaneously optimizes the classification accuracy, feature integrity, and the distribution between source and target. This comprehensive optimization ensures that the model learns robust, domain-invariant features that generalize well across different regions. The proposed method demonstrates strong performance in predicting crop types in regions without labeled data, providing a powerful and adaptable tool for global agricultural monitoring and decision-making. Additionally, we demonstrate how the accuracy improves compared to direct transfer learning-based approaches, showcasing the superiority of domain adaptation strategies in crop-type mapping using remote sensing.

In this paper, we study the performance of few-shot learning, specifically meta learning empowered few-shot relation networks, over supervised deep learning and conventional machine learning approaches in the problem of Sound Source Distance Estimation (SSDE). In previous research on deep supervised SSDE, low accuracies have often resulted from the mismatch between the training data (from known environments) and the test data (from unknown environments). By performing comparative experiments on a sufficient amount of data, we show that the few-shot relation network outperforms other competitors including eXtreme Gradient Boosting (XGBoost), Support Vector Machine (SVM), Convolutional Neural Network (CNN), and MultiLayer Perceptron (MLP). Hence it is possible to calibrate a microphone-equipped system, with a few labeled samples of audio recorded in a particular unknown environment to adjust and generalize our classifier to the possible input data and gain higher accuracies.

OBJECTIVE: This paper aims to address the challenges in abstract screening within systematic reviews (SR) by leveraging the zero-shot capabilities of large language models (LLMs). METHODS: We employ LLM to prioritize candidate studies by aligning abstracts with the selection criteria outlined in an SR protocol. Abstract screening was transformed into a novel question-answering (QA) framework, treating each selection criterion as a question addressed by LLM. The framework involves breaking down the selection criteria into multiple questions, properly prompting LLM to answer each question, scoring and re-ranking each answer, and combining the responses to make nuanced inclusion or exclusion decisions. RESULTS AND DISCUSSION: Large-scale validation was performed on the benchmark of CLEF eHealth 2019 Task 2: Technology-Assisted Reviews in Empirical Medicine. Focusing on GPT-3.5 as a case study, the proposed QA framework consistently exhibited a clear advantage over traditional information retrieval approaches and bespoke BERT-family models that were fine-tuned for prioritizing candidate studies (ie, from the BERT to PubMedBERT) across 31 datasets of 4 categories of SRs, underscoring their high potential in facilitating abstract screening. The experiments also showcased the viability of using selection criteria as a query for reference prioritization. The experiments also showcased the viability of the framework using different LLMs. CONCLUSION: Investigation justified the indispensable value of leveraging selection criteria to improve the performance of automated abstract screening. LLMs demonstrated proficiency in prioritizing candidate studies for abstract screening using the proposed QA framework. Significant performance improvements were obtained by re-ranking answers using the semantic alignment between abstracts and selection criteria. This further highlighted the pertinence of utilizing selection criteria to enhance abstract screening.

The goal of textual adversarial attack methods is to replace some words in an input text in order to make the victim model misbehave. This article proposes an effective word-level adversarial attack method based on sememes and an improved quantum-behaved particle swarm optimization (QPSO) algorithm. The sememe-based substitute method, which uses the words sharing the same sememes as the substitutes of the original words, is first employed to form the reduced search space. Then, an improved QPSO algorithm, called historical information-guided QPSO with random drift local attractor (HIQPSO-RD), is proposed to search the reduced search space for adversarial examples. The HIQPSO-RD introduces historical information into the current mean best position of the QPSO, for the purpose of improving the convergence speed of the algorithm, by enhancing its exploration ability and preventing the premature convergence of the swarm. The proposed algorithm uses the random drift local attractor technique to make a good balance between its exploration and exploitation, so that the algorithm can find a better adversarial attack example with low grammaticality and perplexity (PPL). In addition, it employs a two-stage diversity control strategy to enhance the search performance of the algorithm. Experiments on three natural language processing (NLP) datasets, with three commonly used nature language processing models as victim models, show that our method achieves higher attack success rates but lower modification rates than the state-of-the-art adversarial attack methods. Moreover, the results of human evaluations show that adversarial examples generated by our method can better maintain the semantic similarity and grammatical correctness of the original input.

Hand measurement is vital for hand-centric applications such as glove design, immobilization design, protective gear design, to name a few. Vision-based methods have been previously proposed but are limited in their ability to only extract hand dimensions in a static and standardized posture (open-palm hand). However, dynamic hand measurements should be considered when designing these wearable products since the interaction between hands and products cannot be ignored. Unfortunately, none of the existing methods are designed for measuring dynamic hands. To address this problem, we propose a user-friendly and fast method dubbed Measure4DHand, which automatically extracts dynamic hand measurements from a sequence of depth images captured by a single depth camera. Firstly, the ten dimensions of the hand are defined. Secondly, a deep neural network is developed to predict landmark sequences for the ten dimensions from partial point cloud sequences. Finally, a method is designed to calculate dimension values from landmark sequences. A novel synthetic dataset consisting of 234K hands in various shapes and poses, along with their corresponding ground truth landmarks, is proposed for training the proposed methods. The experiment based on real-world data captured by a Kinect illustrates the evolution of the ten dimensions during hand movement, while the mean ranges of variation are also reported, providing valuable information for the hand wearable product design. (The video abstract is available here.)

In the midst of the COVID-19 pandemic, it was essential to accurately forecast the demand for hospitalisation resources to achieve an effective allocation of healthcare resources. This paper explores the potential of various Deep Learning (DL) models, namely basic Recurrent Neural Networks (RNNs), Long Short-Term Memory networks (LSTMs), Gated Recurrent Units (GRU), Bidirectional RNNs, and Sequence-to-Sequence architectures with the inclusion of attention mechanisms, to forecast the demand for hospitalisation resources (mechanical ventilators) in England during the COVID-19 pandemic. The implementation of simulated annealing (SA) as a hyperparameter tuning method produced certain model structures and good results in terms of prediction accuracy. Our findings show that the LSTM-based models (LSTM_SA), achieved the lowest mean average error (MAE), outperforming other architectures used in this study. The results of this study show the potential of DL models to forecast the demand for resources and could help inform the distribution of hospitalisation resources in England during the COVID-19 pandemic.

Economic dispatch problems (EDPs) can be reduced to non-convex constrained optimization problems, and most of the population-based algorithms are prone to have problems of premature and falling into local optimum when solving EDPs. Therefore, this paper proposes a hybrid quantum-behaved particle swarm optimization (HQPSO) algorithm to alleviate the above problems. In the HQPSO, the Solis and Wets local search method is used to enhance the local search ability of the QPSO so that the algorithm can find solutions that is close to optimal when the constraints are met, and two evolution operators are proposed and incorporated for the purpose of making a better balance between local search and global search abilities at the later search stage. The performance comparison is made among the HQPSO and the other ten population-based random search methods under two different experimental configurations and four different power systems in terms of solution quality, robustness, and convergence property. The experimental results show that the HQPSO improves the convergence properties of the QPSO and finally obtains the best total generation cost without violating any constraints. In addition, the HQPSO outperforms all the other algorithms on 7 cases of all 8 experimental cases in terms of global best position and mean position, which verifies the effectiveness of the algorithm.

The quantum-behaved particle swarm optimization (QPSO) algorithm, a variant of particle swarm optimization (PSO), has been proven to be an effective tool to solve various of optimization problems. However, like other PSO variants, it often suffers a premature convergence, especially when solving complex optimization problems. Considering this issue, this paper proposes a hybrid QPSO with dynamic grouping searching strategy, named QPSO-DGS. During the search process, the particle swarm is dynamically grouped into two subpopulations, which are assigned to implement the exploration and exploitation search, respectively. In each subpopulation, a comprehensive learning strategy is used for each particle to adjust its personal best position with a certain probability. Besides, a modified opposition-based computation is employed to improve the swarm diversity. The experimental comparison is conducted between the QPSO-DGS and other seven state-of-art PSO variants on the CEC’2013 test suit. The experimental results show that QPSO-DGS has a promising performance in terms of the solution accuracy and the convergence speed on the majority of these test functions, and especially on multimodal problems.

Deep neural networks (DNNs) are currently being deployed as machine learning technology in a wide range of important real-world applications. DNNs consist of a huge number of parameters that require millions of floating-point operations (FLOPs) to be executed both in learning and prediction modes. A more effective method is to implement DNNs in a cloud computing system equipped with centralized servers and data storage sub-systems with high-speed and high-performance computing capabilities. This paper presents an up-to-date survey on current state-of-the-art deployed DNNs for cloud computing. Various DNN complexities associated with different architectures are presented and discussed alongside the necessities of using cloud computing. We also present an extensive overview of different cloud computing platforms for the deployment of DNNs and discuss them in detail. Moreover, DNN applications already deployed in cloud computing systems are reviewed to demonstrate the advantages of using cloud computing for DNNs. The paper emphasizes the challenges of deploying DNNs in cloud computing systems and provides guidance on enhancing current and new deployments.

Emerging data-driven technologies and big data analytics generate and deal with high-dimensional data. Transformation of such data into a low-dimensional feature space brings about numerous benefits, such as a more discriminant feature space, performance enhancement, less computational burden, and facilitating data visualization. This paper proposes a novel dimensionality reduction algorithm based on generative adversarial networks to tackle the issues related to high-dimensional data and common challenges in dimensionality reduction. To this aim, two constraints are defined to preserve the characteristics of the original data while rectifying the data distribution upon transformation. Formulating the transformation as sequential projections, the proposed Constrained Adversarial Dimensionality Reduction (CADR) method finds a set of sequential projection vectors that lead to a feature space in which between-class separability and within-class integrity are satisfied. This is while the transformed data perfectly comply with the pairwise affinity correlation in the original feature space. To evaluate the proposed method, nine advanced dimensionality reduction techniques are employed to enable a comparative study. The experiments are performed on several real-world benchmark datasets in terms of classification accuracy, F-measure, and G-mean. The obtained results show that the CADR could yield classification performance at a satisfactory level and outperforms the other competitors.

Remote Human Activity Recognition (HAR) in a private residential area has a beneficial influence on the elderly population's life, since this group of people require regular monitoring of health conditions. This paper addresses the problem of continuous detection of daily human activities using mm-wave Doppler radar. Unlike most previous research, this work records the data in terms of continuous series of activities rather than individual activities. These series of activities are similar to real-life activity patterns. The Dynamic Time Warping (DTW) algorithm is used for the detection of human activities in the recorded time series of data and compared to other time-series classification methods. DTW requires less amount of labelled data. The input for DTW was provided using three strategies, and the obtained results were compared against each other. The first approach uses the pixel-level data of frames (named UnSup-PLevel). In the other two strategies, a Convolutional Variational Autoencoder (CVAE) is used to extract Un-Supervised Encoded features (UnSup-EnLevel) and Supervised Encoded features (Sup-EnLevel) from the series of Doppler frames. Results demonstrates the superiority of the Sup-EnLevel features over UnSup-EnLevel and UnSup-PLevel strategies. However, the performance of the UnSup-PLevel strategy worked surprisingly well without using annotations.

BACKGROUND: A high-quality docking method tends to yield multifold gains with half pains for the new drug development. Over the past few decades, great efforts have been made for the development of novel docking programs with great efficiency and intriguing accuracy. AutoDock Vina (Vina) is one of these achievements with improved speed and accuracy compared to AutoDock4. Since it was proposed, some of its variants, such as PSOVina and GWOVina, have also been developed. However, for all these docking programs, there is still large room for performance improvement. RESULTS: In this work, we propose a parallel multi-swarm cooperative particle swarm model, in which one master swarm and several slave swarms mutually cooperate and co-evolve. Our experiments show that multi-swarm programs possess better docking robustness than PSOVina. Moreover, the multi-swarm program based on random drift PSO can achieve the best highest accuracy of protein-ligand docking, an outstanding enrichment effect for drug-like activate compounds, and the second best AUC screening accuracy among all the compared docking programs, but with less computation consumption than most of the other docking programs. CONCLUSION: The proposed multi-swarm cooperative model is a novel algorithmic modeling suitable for protein-ligand docking and virtual screening. Owing to the existing coevolution between the master and the slave swarms, this model in parallel generates remarkable docking performance. The source code can be freely downloaded from https://github.com/li-jin-xing/MPSOVina .

In general, flexible ligand docking is used for docking simulations under the premise that the position of the binding site is already known, and meanwhile it can also be used without prior knowledge of the binding site. However, most of the optimization search algorithms used in popular docking software are far from being ideal in the first case, and they can hardly be directly utilized for the latter case due to the relatively large search area. In order to design an algorithm that can flexibly adapt to different sizes of the search area, we propose an effective swarm intelligence optimization algorithm in this paper, called diversity-controlled Lamarckian quantum particle swarm optimization (DCL-QPSO). The highlights of the algorithm are a diversity-controlled strategy and a modified local search method. Integrated with the docking environment of Autodock, the DCL-QPSO is compared with Autodock Vina, Glide and other two Autodock-based search algorithms for flexible ligand docking. Experimental results revealed that the proposed algorithm has a performance comparable to those of Autodock Vina and Glide for dockings within a certain area around the binding sites, and is a more effective solver than all the compared methods for dockings without prior knowledge of the binding sites.

Connected and Autonomous Vehicles (CAV) are gaining increasing importance due to the current needs of modern society for better mobility and societal impact. CAV development and adoption will be driven by Artificial Intelligence (AI) and 5G/6G technologies which will offer increased speed, reduced latency and ubiquity. However, the public is concerned with the concept of handing total control of driving to vehicles. These concerns will inhibit the adoption of CAVs when they become available to the public. In this paper, we investigated user adoption of CAVs by collecting quantitative data from potential users based on their preference and inherent concerns towards adoption. We conducted a statistical analysis and applied machine learning techniques to predict the user adoption for CAVs. Our results show that several machine learning approaches were effective in forecasting user adoption for CAVs. We have employed Neural Networks, Random Forest, Naïve Bayes and Fuzzy Logic based models and achieved accuracies of 81.76%, 83.63%, 82.15% and 86.38, respectively, in forecasting the public adoption of CAV.

As fully autonomous driving is introduced on our roads, the safety of vulnerable road users is of the greatest importance. Available real-world data is limited and often lacks the variety required to ensure the safe deployment of new technologies. This paper builds on a novel generation method to generate pedestrian crossing scenarios for autonomous vehicle testing, known as the Ped-Cross GAN. While our previously developed Pedestrian Scenario dataset [1] is extremely detailed, there exist labels in the dataset where available data is severely imbalanced. In this paper, augmented non-roadside data is used to improve the generation results of pedestrians running at the roadside, increasing the classification accuracy from 20.95% to 82.56%, by increasing the training data by only 30%. This proves that researchers can generate rare, edge case scenarios using the Ped-Cross GAN, by successfully supplementing available data with additional non-roadside data. This will allow for adequate testing and greater test coverage when testing the performance of autonomous vehicles in pedestrian crossing scenarios. Ultimately, this will lead to fewer pedestrian casualties on our roads.

In this paper, a deep learning approach is proposed to accurately position wheeled vehicles in Global Navigation Satellite Systems (GNSS) deprived environments. In the absence of GNSS signals, information on the speed of the wheels of a vehicle (or other robots alike), recorded from the wheel encoder, can be used to provide continuous positioning information for the vehicle, through the integration of the vehicle's linear velocity to displacement. However, the displacement estimation from the wheel speed measurements are characterised by uncertainties, which could be manifested as wheel slips or/and changes to the tyre size or pressure, from wet and muddy road drives or tyres wearing out. As such, we exploit recent advances in deep learning to propose the Wheel Odometry neural Network (WhONet) to learn the uncertainties in the wheel speed measurements needed for correction and accurate positioning. The performance of the proposed WhONet is first evaluated on several challenging driving scenarios, such as on roundabouts, sharp cornering, hard-brake and wet roads (drifts). WhONet's performance is then further and extensively evaluated on longer-term GNSS outage scenarios of 30s, 60s, 120s and 180s duration, respectively over a total distance of 493 km. The experimental results obtained show that the proposed method is able to accurately position the vehicle with up to 93% reduction in the positioning error of its original counterpart after any 180s of travel. WhONet's implementation can be found at https://github.com/onyekpeu/WhONet.

Deep neural networks (NNs) have been proved to be efficient learning systems for supervised and unsupervised tasks. However, learning complex data representations using deep NNs can be difficult due to problems such as lack of data, exploding or vanishing gradients, high computational cost, or incorrect parameter initialization, among others. Deep representation and transfer learning (RTL) can facilitate the learning of data representations by taking advantage of transferable features learned by an NN model in a source domain, and adapting the model to a new domain.

Inertial Navigation Systems (INS) are commonly used to localise vehicles in the absence of Global Navigation Satellite Systems (GNSS) signals. However, they are plagued by noises, which grow exponentially over time during the triple integration computation, leading to a poor navigation solution. We explore the wheel encoder as an alternative to the accelerometer of the INS for positional tracking, and for the first time investigate the capability of deep learning using the Long Short-Term Memory (LSTM) neural network to learn the uncertainty inherent in the wheel speed measurements. These uncertainties could be manifested as changes in the tyre size or pressure, or wheel slips as a result of worn out tyres or wet/muddy road drive. The proposed solution has less integration steps in its computation, therefore providing the potential for a more accurate positioning estimation. Through a performance evaluation on several challenging scenarios for vehicular driving, such as hard braking, quick changes in vehicular acceleration, and wet/muddy road driving, we show that the wheel speed-based positioning approach is able to achieve up to 81.46 % improvement compared to the INS accelerometer approach.

In this work, we propose a novel Generative Adversarial Stacked Autoencoder that learns to map facial expressions with up to ±60 degrees to an illumination invariant facial representation of 0 degrees. We accomplish this by using a novel convolutional layer that exploits both local and global spatial information, and a convolutional layer with a reduced number of parameters that exploits facial symmetry. Furthermore, we introduce a generative adversarial gradual greedy layer-wise learning algorithm designed to train Adversarial Autoencoders in an efficient and incremental manner. We demonstrate the efficiency of our method and report state-of-the-art performance on several facial emotion recognition corpora, including one collected in the wild.

Deep neural networks have proven to be efficient systems for learning complex data representations. However, one of their main constraints is their inability to deal with changes in the data distribution. For instance, in real-time facial expression recognition, the data used to evaluate a model commonly differs in quality compared to that used to train the model, leading to poor generalization performance. In this work we propose a novel Deep Convolutional Neural Network (CNN) architecture pre-trained as a Stacked Convolutional Autoencoder (SCAE) to address emotion recognition in unconstrained environments. The SCAE is trained in a greedy layer-wise unsupervised fashion, and combines convolutional and fully connected layers and learns to encode facial expression images as an illumination and facial pose invariant feature vector. The CNN offers state-of-the-art classification rate of 99.52% on a combined corpus of gamma corrected version of the CK+, JAFFE, FEEDTUM and KDEF datasets. When evaluated on unseen data obtained in unconstrained environments, our approach achieves 79.75%, an increase of over 28% compared to a CNN without our pre-training approach, supporting the methodology proposed in this work.

BACKGROUND: Protein-ligand docking has emerged as a particularly important tool in drug design and development, and flexible ligand docking is a widely used method for docking simulations. Many docking software packages can simulate flexible ligand docking, and among them, Autodock is widely used. Focusing on the search algorithm used in Autodock, many new optimization approaches have been proposed over the last few decades. However, despite the large number of alternatives, we are still lacking a search method with high robustness and high performance. RESULTS: In this paper, in conjunction with the popular Autodock software, a novel hybrid version of the random drift particle swarm optimization (RDPSO) algorithm, called diversity-guided Lamarckian RDPSO (DGLRDPSO), is proposed to further enhance the performance and robustness of flexible ligand docking. In this algorithm, a novel two-phase diversity control (2PDC) strategy and an efficient local search strategy are used to improve the search ability and robustness of the RDPSO algorithm. By using the PDBbind coreset v.2016 and 24 complexes with apo-structures, the DGLRDPSO algorithm is compared with the Lamarckian genetic algorithm (LGA), Lamarckian particle swarm optimization (LPSO) and Lamarckian random drift particle swarm optimization (LRDPSO). The experimental results show that the 2PDC strategy is able to enhance the robustness and search performance of the proposed algorithm; for test cases with different numbers of torsions, the DGLRDPSO outperforms the LGA and LPSO in finding both low-energy and small-RMSD docking conformations with high robustness in most cases. CONCLUSION: The DGLRDPSO algorithm has good search performance and a high possibility of finding a conformation with both a low binding free energy and a small RMSD. Among all the tested algorithms, DGLRDPSO has the best robustness in solving both holo- and apo-structure docking problems with different numbers of torsions, which indicates that the proposed algorithm is a reliable choice for the flexible ligand docking in Autodock software.

The exponential growth in the production of video contents in different industries causes an urgent need for effective video summarization (VS) techniques, in order to get an optimal storage and preservation of key information in the video. Compared to other domains, industrial videos are more challenging to process, as they usually contain diverse and complex events, which make their online processing a difficult task. In this article, we introduce an online system for intelligent video capturing, coarse and fine redundancy removal, and summary generation. First, we capture video data through resource-constrained devices in an industrial Internet of Things network, equipped with vision sensors and apply coarse redundancy removal through the comparison of low-level features. Second, we transmit the resulting frames to the cloud for detailed analysis, where sequential features are extracted for the selection of candidate keyframes. Finally, we refine the candidate keyframes in order to discriminate those with maximum information as part of the summary. The key contributions of this article include the coarse and fine refining of video data implemented over resource-restricted devices and the presentation of important data in the form of a summary. Experiments11[Online]. Available: https://github.com/tanveer-hussain/DeepRes-Video-Summarization. over publicly available datasets evince a 0.3-unit increase in the F1 score when compared to state-of-the-art and with reduced time complexity. Furthermore, we provide convincing results on our newly created dataset in an industrial environment, which is made publicly available for the research community along with its labeled ground truth.

Smoke detection in foggy surveillance environments is a challenging task and plays a key role in disaster management for industrial systems. The current smoke detection methods are applicable to only normal surveillance videos, providing unsatisfactory results for video streams captured from foggy environments, due to challenges related to clutter and unclear contents. In this paper, an energy-friendly edge intelligence-assisted smoke detection method is proposed using deep convolutional neural networks for foggy surveillance environments. Our method uses a light-weight architecture, considering all necessary requirements regarding accuracy, running time, and deployment feasibility for smoke detection in an industrial setting, compared to other complex and computationally expensive architectures including AlexNet, GoogleNet, and visual geometry group (VGG). Experiments are conducted on available benchmark smoke detection datasets, and the obtained results show better performance of the proposed method over state-of-the-art for early smoke detection in foggy surveillance.

There has been a growing interest in using smart grids due to their capability in delivering automated and distributed energy level to the consumption units. However, in order to guarantee the safe and reliable delivery of the high-quality power from the generation units to the consumers, smart grids need to be equipped with diagnostic systems. This paper presents an efficient data-driven scheme for diagnosing faults in smart grids. In order to reduce the computational burden and monitor the state of the system with a lower number of smart meters, a method based on the affinity propagation clustering algorithm is suggested for the placement of meters, that makes use of the graph-based representation of the system. The collected voltage data measurements from the installed meters are then decomposed by matching pursuit decomposition in order to generate informative features. Extracted features are then used to train a convolutional neural network, and the constructed deep learning model is then tested using unseen samples of normal and faulty conditions. Simulation results based on the IEEE 39-Bus System demonstrate the effectiveness of the proposed data-driven fault diagnostic system.

This paper describes the core computational mechanisms used by an urban flood forecasting and monitoring platform developed as part of a UK Newton Fund project in Malaysia. FLUD-FLood monitoring and forecasting platform for Urban Deployment - is a novel system aiming to deliver an effective and low cost urban flood forecasting solution, which is able to accurately forecast flood risk at street level, and deliver optimized recommendations to the relevant authorities as well as an early warning alerts to members of the public. This platform is based on a hybrid Deep Learning and Fuzzy Logic based architecture. As demonstrated by the experimental results and the analysis presented in this paper, this architecture enables the proposed system to account for factors that are not included in other modern flood forecasting systems, and simultaneously process high volumes of data originating from diverse data sources, in order to deliver accurate predictions concerning urban flood events.

The safety of vulnerable road users is of paramount importance as transport moves towards fully automated driving. The richness of real-world data required for testing autonomous vehicles is limited, and furthermore, the available data does not have a fair representation of different scenarios and rare events. This work presents a novel approach for the generation of human pose structures, specifically the type of pose structures that would appear to be in pedestrian scenarios. The results show that the generated pedestrian structures are indistinguishable from the ground truth pose structures when classified using a suitably trained classifier. The paper demonstrates that the Generative Adversarial Network architecture can be used to create realistic new training samples, and, in future, new pedestrian events.

Text contained within figures is an important source of information in biomedical literature. Despite this, end-to-end text extraction from biomedical figures remains a challenging task. This paper presents a novel approach to address the founding block of this task, text detection, not only from biomedical figures but also from images in general. Particularly, the paper proposes an approach that simplifies the text detection problem into a reconstruction problem using a deep convolutional neural network. Designed to overcome the specific challenges of text detection from biomedical figures, our proposed model reports promising results on the DETEXT dataset.

The domain of automatic face detection is a challenging problem that has made great progress in the last two decades. Much of this progress is due to the continuous advancements in deep learning and computer vision. Nonetheless, contemporary state-of-the-art face detection models rely on exhaustive search or are unable to deal with changes in the data distribution. In this work, we propose a novel Deep Reinforcement Learning (DRL) approach for face detection on data with nonuniform conditions. More specifically, we address illumination and rotation invariance in face detection. Firstly, we train a Stacked Convolutional Autoencoder (SCAE) in a greedy layer-wise unsupervised fashion for illumination invariant feature extraction. We then train a deep Q-network on the illumination invariant features produced by the SCAE model, to learn an action-value policy that allows an agent to place a bounding box around a face. The proposed approach achieves state-of-the-art recognition rates on images with varying degrees of illumination and images that contain faces with some degree of rotation.

Recent advances in automatic machine learning (aML) allow solving problems without any human intervention. However, sometimes a human-in-the-loop can be beneficial in solving computationally hard problems. In this paper we provide new experimental insights on how we can improve computational intelligence by complementing it with human intelligence in an interactive machine learning approach (iML). For this purpose, we used the Ant Colony Optimization (ACO) framework, because this fosters multi-agent approaches with human agents in the loop. We propose unification between the human intelligence and interaction skills and the computational power of an artificial system. The ACO framework is used on a case study solving the Traveling Salesman Problem, because of its many practical implications, e.g. in the medical domain. We used ACO due to the fact that it is one of the best algorithms used in many applied intelligence problems. For the evaluation we used gamification, i.e. we implemented a snake-like game called Traveling Snakesman with the MAX–MIN Ant System (MMAS) in the background. We extended the MMAS–Algorithm in a way, that the human can directly interact and influence the ants. This is done by “traveling” with the snake across the graph. Each time the human travels over an ant, the current pheromone value of the edge is multiplied by 5. This manipulation has an impact on the ant’s behavior (the probability that this edge is taken by the ant increases). The results show that the humans performing one tour through the graphs have a significant impact on the shortest path found by the MMAS. Consequently, our experiment demonstrates that in our case human intelligence can positively influence machine intelligence. To the best of our knowledge this is the first study of this kind.

Abstract This paper describes a roadmap for the development of the SP Machine, based on the SP Theory of Intelligence and its realization in the SP Computer Model. The SP Machine will be developed initially as a software virtual machine with high levels of parallel processing, hosted on a high-performance computer. The system should help users visualize knowledge structures and processing. Research is needed into how the system may discover low-level features in speech and in images. Strengths of the SP System in the processing of natural language may be augmented, in conjunction with the further development of the SP System’s strengths in unsupervised learning. Strengths of the SP System in pattern recognition may be developed for computer vision. Work is needed on the representation of numbers and the performance of arithmetic processes. A computer model is needed of SP-Neural, the version of the SP Theory expressed in terms of neurons and their interconnections. The SP Machine has potential in many areas of application, several of which may be realized on short-to-medium timescales.

Employing effective plagiarism detection methods are seen to be essential in the next generation web. In this paper, we present a novel approach for plagiarism detection without reference collections. The proposed approach relies on using some statistical properties of the most common words, and the Latent Semantic Analysis that is applied to extract the most common words usage patterns. This method aims to generate a model of author’s “style” by revealing a set of certain features of authorship. The model generation procedure focuses on just one author, as an attempt to summarise the aspects of an author’s style in a definitive and clear-cut manner. The feature set of the intrinsic model were based on the frequency of the most common words, their relative frequencies in the book series, and the deviation of these frequencies across all books for a particular author. The approach has been evaluated using the leave-one-out-cross-validation method on the CEN (Corpus of English Novel) data set. Results have indicated that, by integrating deep latent semantic and stylometric analyses, hidden changes can be identified when a reference collection does not exist. The results have also shown that our Multi-Layer Perceptron based approach statistically outperforms Bayesian Network, Support Vector Machine and Random Forest models, by accurately predicting the author classes with an overall accuracy of 97%.

In order to effectively identify and classify weld defects of thin-walled metal canisters, a weld defect detection and classification algorithm based on machine vision is proposed in this paper. With the weld defects categorized, a modified background subtraction method based on Gaussian mixture models, is proposed to extract the feature areas of the weld defects. Then, we design an algorithm for weld detection and classification according to the extracted features. Next, by using the weld images sampled by the constructed weld defect detection system on a real-world production line, the parameters of the weld defect classifiers are determined empirically. Experimental results show that the proposed methods can identify and classify the weld defects with more than 95% accuracy rate. Moreover, the weld detection results obtained in the actual production line show that the detection and classification accuracy can reach more than 99%, which means that the system enhanced with the proposed method can meet the requirements for the best real-time and continuous weld defect detection systems available nowadays.

Fingerprints, as one of the most widely used biometric modalities, can be used to identify and distinguish between genders. Gender classification is very important in reducing the time when investigating criminal offenders and gender impersonation. In this work, we use deep Convolutional Neural Networks (CNNs) to not only classify fingerprints by gender, but also identify individual hands and fingers. Transfer learning is employed to speed up the training of the CNN. The CNN achieves an accuracy of 75.2%, 93.5%, and 76.72% for the classification of gender, hand, and fingers, respectively. These results obtained using our publicly available Sokoto Coventry Fingerprint Dataset (SOCOFing) serve as benchmark classification results on this dataset.

In this work we propose a novel method to address illumination invariance for facial expression recognition. We propose a Deep Convolutional Network (CNN) pre-trained as a Deep Stacked Convolutional Autoencoder (SCAE) in a greedy layer-wise unsupervised fashion. The SCAE model learns to encode facial expression images and produce a feature vector with relatively similar illumination, regardless of the luminance level of the input image. Moreover, we propose fine-tuning the stacked shallow autoencoders after each one of these is trained greedily, rather than just at the end, and show that this approach significantly improves the set of illumination invariant features learnt by the SCAE. Finally, we propose the use of a variant rectifier linear unit transfer function that helps the SCAE model reduce or increase the illumination of images with high or low luminance, and show that the lower and upper bounds greatly influence classification performance. The method proposed provides an increase in classification accuracy of 4% on the KDEF dataset and 8% on the CK+ dataset.

Using short URLs in Twitter messages has increased in popularity in the past few years. This is mostly due to the fact that Twitter, as one of the most popular social media networks, imposes a 140 character limit to the messages distributed over the network. This paper analyzes the use of short URLs by Twitter users. Specifically, the goal is to examine the content pointed by the short URLs as well as the potential impact on the performance of sentiment analysis (opinion mining) tasks. Opinion mining based on Twitter feed has been used in an array of applications, including healthcare, identifying public opinion on political issues, financial modeling and advertising. Past research has however completely disregarded tweets which contain URLs. It is not hard to see how opinion mining can be improved considering the fact that Twitter users regularly post URLs pointing to articles endorsing a particular political figure, articles in important financial outlets or reviews of products. This study is based on the analysis of three distinct Twitter datasets with varying number of tweets which include short URLs. Popular machine learning techniques used in opinion mining were deployed in different experimental settings to conclude which are the most lucrative options.

This paper proposes an adaptive incremental ensemble of extreme learning machines for fault diagnosis. The diagnostic system contains a data processing unit which aims to progressively generate discriminant features from the vibration signals for decision making. The decision making unit receives a few sets of labeled discriminant features in a chunk by chunk manner, incrementally learns the features-faults relations, dynamically diagnoses multiple bearing defects, and adaptively adjusts itself to learn new concept classes. This adaptive ensemble system is based on incremental learning of multiple extreme learning machines that are able to consult together and adjust themselves based on their confidence in the decision making. Extreme learning machines are used to construct the hybrid ensemble due to their good controllability and fast learning rate. Experimental results show the efficiency of the hybrid diagnostic system. The proposed diagnostic system is applied to diagnosing bearing defects in an induction motor.

Emotion recognition is critical for everyday living and is essential for meaningful interaction. If we are to progress towards human and machine interaction that is engaging the human user, the machine should be able to recognize the emotional state of the user. Deep Convolutional Neural Networks (CNN) have proven to be efficient in emotion recognition problems. The good degree of performance achieved by these classifiers can be attributed to their ability to self-learn a down-sampled feature vector that retains spatial information through filter kernels in convolutional layers. Given the view that random initialization of weights can lead to convergence to non-optimal local minima, in this paper we explore the impact of training the initial weights in an unsupervised manner. We study the effect of pre-training a Deep CNN as a Stacked Convolutional Auto-Encoder (SCAE) in a greedy layer-wise unsupervised fashion for emotion recognition using facial expression images. When trained with randomly initialized weights, our CNN emotion recognition model achieves a performance rate of 91.16% on the Karolinska Directed Emotional Faces (KDEF) dataset. In contrast, when each layer of the model, including the hidden layer, is pre-trained as an Auto-Encoder, the performance increases to 92.52%. Pre-training our CNN as a SCAE also reduces training time marginally. The emotion recognition model developed in this work will form the basis of a real-time empathic robot system.

The papers in this special section focus on the use of fuzzy techniques and logic for use in financial modeling and simulation. Computational intelligence has attracted a significant and increasing interest from the financial engineering community, and an emerging interest from analytical economics groups. The bar has been raised with the revision of regulations, and the required compliance and risk management. The new rules should be implemented through new processes and supported by developing new computational tools. Computational systems, capturing sentiments, preferences, behavior, and beliefs, are becoming indispensable in financial applications and desirable in economic analysis. They address problems in the classification of credit worthiness and fraud detection, contribute to the analysis and pricing of financial instruments, and effectively support portfolio optimization and investment analysis. They are instrumental in the design of market mechanisms and contagion mechanisms, and are contributing to the simulation of micro- and macro-economic processes. The armory of fuzzy techniques is capable of addressing challenges encountered in financial engineering and analytical economics. Fuzzy logic can effectively describe and incorporate experts’ intuition, market participants’ preferences, and economic agents’ behavior, thus reaching beyond the capabilities of probabilistic models. The objective of this special issue is to bring together the most recent advances in the design and application of fuzzy approaches to real problems in financial engineering and analytical economics.

Plagiarism detection is gaining increasing importance due to requirements for integrity in education. In this paper, we have developed a new integrated approach for extrinsic plagiarism detection. The proposed approach is based on four well-known models namely Bag of Words (BOW), Latent Semantic Analysis (LSA), Stylometry and Support Vector Machines (SVM). The proposed approach works by capturing usage patterns of the most common words (MCW) from books of 25 authors. Stylistic features for each author were harnessed in the method by adjusting the LSA weighting technique. The adjusted LSA method was trained in a novel manner using the leave-one-out-cross-validation technique and compared with the traditional LSA method. The results have shown that the enhanced weighting method of the adjusted LSA outperforms the traditional LSA method.

Low-rank representation (LRR) and its variations have recently attracted a great deal of attention because of its effectiveness in exploring low-dimensional subspace structures embedded in data. LRR-related algorithms have many applications in computer vision, signal processing, semi-supervised learning and pattern recognition. However, most of the existing LRR methods fail to take into account the non-linear geometric structures within data, thus the locality and the similarity information among data may be missing in the learning process, which have been shown to be beneficial for discriminative tasks. To improve LRR in this regard, we propose a manifold locality constrained low-rank representation framework (MLCLRR) for data representation. By taking the local manifold structure of the data into consideration, the proposed MLCLRR method not only can represent the global low-dimensional structures, but also capture the local intrinsic non-linear geometric information in the data. The experimental results on different types of vision problems demonstrate the effectiveness of the proposed method.

In this paper, a supervised fuzzy-neighborhood density-based clustering approach is proposed for the fault diagnosis of induction motors' bearings. The proposed approach makes use of the labeled data regarding the actual classes of faulty and fault-free cases, in order to train the fuzzy-neighborhood density-based clustering algorithm in a supervised manner, by resorting to an invasive weed optimization algorithm that aims to minimize an error-based objective function. The proposed classifier can properly classify multi-class data with complex and variously shaped decision boundaries among the different classes of faults and the fault-free state, and is robust against noise. This is due mainly to the fact that the classifier is constructed using the fuzzy-neighborhood density based clustering method, which is not sensitive to the geometrical shape of clusters in the feature space.

Proper detection of unknown patterns plays an important role in diagnosing new classes of faults. This can be done by incremental learning of novel information and updating the diagnostic system by appending newly trained fault classifiers in an ensemble design. We consider a new-class fault detector previously developed by the authors and based on thresholding the normalized weighted average of the outputs (NWAO) of the base classifiers in a multi-classifier diagnostic system. A proper tuning of the thresholds in the NWAO detector is necessary to achieve a satisfactory performance. This is done in this paper by specifically introducing a performance function and optimizing it within the necessary trade-off between new class false alarm and new class missed alarm rates, by means of an Invasive Weed Optimization (IWO) algorithm. The optimal NWAO detector is tested with respect to a set of simulated sensor faults in the doubly-fed induction generator (DFIG) of a wind turbine.

BACKGROUND: Finding an efficient method to solve the parameter estimation problem (inverse problem) for nonlinear biochemical dynamical systems could help promote the functional understanding at the system level for signalling pathways. The problem is stated as a data-driven nonlinear regression problem, which is converted into a nonlinear programming problem with many nonlinear differential and algebraic constraints. Due to the typical ill conditioning and multimodality nature of the problem, it is in general difficult for gradient-based local optimization methods to obtain satisfactory solutions. To surmount this limitation, many stochastic optimization methods have been employed to find the global solution of the problem. RESULTS: This paper presents an effective search strategy for a particle swarm optimization (PSO) algorithm that enhances the ability of the algorithm for estimating the parameters of complex dynamic biochemical pathways. The proposed algorithm is a new variant of random drift particle swarm optimization (RDPSO), which is used to solve the above mentioned inverse problem and compared with other well known stochastic optimization methods. Two case studies on estimating the parameters of two nonlinear biochemical dynamic models have been taken as benchmarks, under both the noise-free and noisy simulation data scenarios. CONCLUSIONS: The experimental results show that the novel variant of RDPSO algorithm is able to successfully solve the problem and obtain solutions of better quality than other global optimization methods used for finding the solution to the inverse problems in this study.

Hidden Markov Models (HMMs) are powerful tools for multiple sequence alignment (MSA), which is known to be an NP-complete and important problem in bioinformatics. Learning HMMs is a difficult task, and many meta-heuristic methods, including particle swarm optimization (PSO), have been used for that. In this paper, a new variant of PSO, called the random drift particle swarm optimization (RDPSO) algorithm, is proposed to be used for HMM learning tasks in MSA problems. The proposed RDPSO algorithm, inspired by the free electron model in metal conductors in an external electric field, employs a novel set of evolution equations that can enhance the global search ability of the algorithm. Moreover, in order to further enhance the algorithmic performance of the RDPSO, we incorporate a diversity control method into the algorithm and, thus, propose an RDPSO with diversity-guided search (RDPSO-DGS). The performances of the RDPSO, RDPSO-DGS and other algorithms are tested and compared by learning HMMs for MSA on two well-known benchmark data sets. The experimental results show that the HMMs learned by the RDPSO and RDPSO-DGS are able to generate better alignments for the benchmark data sets than other most commonly used HMM learning methods, such as the Baum-Welch and other PSO algorithms. The performance comparison with well-known MSA programs, such as ClustalW and MAFFT, also shows that the proposed methods have advantages in multiple sequence alignment.

This paper proposes the random drift particle swarm optimization (RDPSO) algorithm to solve economic dispatch (ED) problems from power systems area. The RDPSO is inspired by the free electron model in metal conductors placed in an external electric field, and it employs a novel set of evolution equations that can enhance the global search ability of the algorithm. Many nonlinear characteristics of a power generator, such as the ramp rate limits, prohibited operating zones and nonsmooth cost functions are considered when the proposed method is used in practice for optimizing the generators' operation. The performance of the RDPSO method is evaluated on three different power systems, and compared with that of other optimization methods in terms of the solution quality, robustness, and convergence performance. The experimental results show that the RDPSO method performs better in solving the ED problems than any other tested optimization techniques.

This paper introduces a variant of particle swarm optimization algorithm called the drift particle swarm optimization (DPSO), which is inspired by the free electron model in an external electric field at finite temperature. As the compression-expansion coefficient in DPSO is an important parameter which can greatly influence the performance of the algorithm, three types of control strategies are proposed to control this parameter. The performance of these strategies on the DPSO is comprehensively evaluated on eight benchmark functions. From the experimental results and statistical tests, guidelines about selecting the control method for the compression-expansion coefficient are given.

Quantum-behaved particle swarm optimization (QPSO), motivated by concepts from quantum mechanics and particle swarm optimization (PSO), is a probabilistic optimization algorithm belonging to the bare-bones PSO family. Although it has been shown to perform well in finding the optimal solutions for many optimization problems, there has so far been little analysis on how it works in detail. This paper presents a comprehensive analysis of the QPSO algorithm. In the theoretical analysis, we analyze the behavior of a single particle in QPSO in terms of probability measure. Since the particle's behavior is influenced by the contraction-expansion (CE) coefficient, which is the most important parameter of the algorithm, the goal of the theoretical analysis is to find out the upper bound of the CE coefficient, within which the value of the CE coefficient selected can guarantee the convergence or boundedness of the particle's position. In the experimental analysis, the theoretical results are first validated by stochastic simulations for the particle's behavior. Then, based on the derived upper bound of the CE coefficient, we perform empirical studies on a suite of well-known benchmark functions to show how to control and select the value of the CE coefficient, in order to obtain generally good algorithmic performance in real world applications. Finally, a further performance comparison between QPSO and other variants of PSO on the benchmarks is made to show the efficiency of the QPSO algorithm with the proposed parameter control and selection methods.

Abstract In plagiarism detection (PD) systems, two important problems should be considered: the problem of retrieving candidate documents that are globally similar to a document q under investigation, and the problem of side‐by‐side comparison of q and its candidates to pinpoint plagiarized fragments in detail. In this article, the authors investigate the usage of structural information of scientific publications in both problems, and the consideration of citation evidence in the second problem. Three statistical measures namely Inverse Generic Class Frequency, Spread, and Depth are introduced to assign a degree of importance (i.e., weight) to structural components in scientific articles. A term‐weighting scheme is adjusted to incorporate component‐weight factors, which is used to improve the retrieval of potential sources of plagiarism. A plagiarism screening process is applied based on a measure of resemblance, in which component‐weight factors are exploited to ignore less or nonsignificant plagiarism cases. Using the notion of citation evidence, parts with proper citation evidence are excluded, and remaining cases are suspected and used to calculate the similarity index. The authors compare their approach to two flat‐based baselines, TF‐IDF weighting with a Cosine coefficient, and shingling with a Jaccard coefficient. In both baselines, they use different comparison units with overlapping measures for plagiarism screening. They conducted extensive experiments using a dataset of 15,412 documents divided into 8,657 source publications and 6,755 suspicious queries, which included 18,147 plagiarism cases inserted automatically. Component‐weight factors are assessed using precision, recall, and F‐measure averaged over a 10‐fold cross‐validation and compared using the ANOVA statistical test. Results from structural‐based candidate retrieval and plagiarism detection are evaluated statistically against the flat baselines using paired‐t tests on 10‐fold cross‐validation runs, which demonstrate the efficacy achieved by the proposed framework. An empirical study on the system's response shows that structural information, unlike existing plagiarism detectors, helps to flag significant plagiarism cases, improve the similarity index, and provide human‐like plagiarism screening results.

Over the past two decades, neural networks have been applied to develop short-term traffic flow predictors. The past traffic flow data, captured by on-road sensors, is used as input patterns of neural networks to forecast future traffic flow conditions. The amount of input patterns captured by the on-road sensors is usually huge, but not all input patterns are useful when trying to predict the future traffic flow. The inclusion of useless input patterns is not effective to developing neural network models. Therefore, the selection of appropriate input patterns, which are significant for short-term traffic flow forecasting, is essential. This can be conducted by setting an appropriate configuration of input nodes of the neural network; however, this is usually conducted by trial and error. In this paper, the Taguchi method, which is a robust and systematic optimization approach for designing reliable and high-quality models, is proposed for the purpose of determining an appropriate neural network configuration, in terms of input nodes, in order to capture useful input patterns for traffic flow forecasting. The effectiveness of the Taguchi method is demonstrated by a case study, which aims to develop a short-term traffic flow predictor based on past traffic flow data captured by on-road sensors located on a Western Australia freeway. Three advantages of using the Taguchi method were demonstrated: 1) short-term traffic flow predictors with high accuracy can be designed; 2) the development time for short-term traffic flow predictors is reasonable; and 3) the accuracy of short-term traffic flow predictors is robust with respect to the initial settings of the neural network parameters during the learning phase.

We believe the great interpretability of fuzzy models allow fuzzy-based methods to play a very important role in Microarray gene expression data analysis, but the advantages offered by fuzzy-based techniques in this application have not yet been fully explored in the literature. In this paper, we construct Multi-Objective Evolutionary Algorithms based Interpretable Fuzzy (MOEAIF) models for microarray gene expression data analysis. Our novel fuzzy models can significantly decrease the model complexity, and automatically balance the accuracy and interpretability of the models. The experimental studies have shown that relatively simple and small fuzzy rule bases, with satisfactory classification performance, have been successful found for challenging microarray gene expression datasets.

Summary form only given. Machine Learning has become a very popular approach in addressing problems in the Computational Biology and Bioinformatics area. In addition, multi-classifier systems have also gained popularity among researchers working in machine learning and applications for their ability to fuse together multiple models and obtain better overall accuracy and classification results. This talk is concerned with current issues in the design of multi-classifier systems and presents some multi-classifier developments for several bioinformatics problems. The talk will first present an overview and current status of machine learning methods in bioinformatics and computational biology. The talk will then bring in some important issues in building ensembles of classifiers, with a focus on the diversity and combination of individual classifiers. Few diversification and combination schemes are presented along with guidelines for the selection of different training paradigms and performance metrics, based on the properties and distribution of the data. Then, the presentation will proceed with introducing our computational intelligence based multi-classifier developments for solving several bioinformatics problems, such as recognizing sequences in DNA strings, micro-array gene expression data analysis, protein structure prediction. The talk will also present related machine learning issues in developing such systems, such as learning from imbalanced datasets and using appropriate performance metrics for model selection and evaluation. The presented approaches and results will advocate that ensembles of classifiers can be used as effective modelling tools in solving challenging bioinformatics problems.

Networks are very important in many fields of machine learning research. Within networks research, inferring the structure of unknown networks is often a key problem; e.g. of genetic regulatory networks. However, there are very few well-known biological networks, and good simulation is essential for validating and evaluating novel structural inference techniques. Further, the importance of large, genome-wide structural inference is increasingly recognised, but there does not appear to be a good simulator available for large networks. This paper presents GreenSim, a simulator that helps address this gap. GreenSim automatically generates large, genome-size networks with more biologically realistic structural characteristics and 2nd-order non-linear regulatory functions. The simulator itself and the novel method used for generating a network structure with appropriate in- and out-degree distributions may also generalise easily to other types of network. GreenSim is available online at: http://syntilect.com/cgf/pubs:software.

This paper presents a novel algorithm for multiobjective training of Radial Basis Function (RBF) networks based on least-squares and Particle Swarm Optimization methods. The formulation is based on the fundamental concept that supervised learning is a bi-objective optimization problem, in which two conflicting objectives should be minimized. The objectives are related to the empirical training error and the machine complexity. The training is done in three steps: i) a conventional minimization of the training error, ii) multiobjective least-squares optimization for the linear parameters and, iii) particle swarm optimization for the nonlinear parameters. Some results are presented and they show the effectiveness of the proposed approach.

Random undersampling and oversampling are simple but well-known resampling methods applied to solve the problem of class imbalance. In this paper we show that the random oversampling method can produce better classification results than the random undersampling method, since the oversampling can increase the minority class recognition rate by sacrificing less amount of majority class recognition rate than the undersampling method. However, the random oversampling method would increase the computational cost associated with the SVM training largely due to the addition of new training examples. In this paper we present an investigation carried out to develop efficient resampling methods that can produce comparable classification results to the random oversampling results, but with the use of less amount of data. The main idea of the proposed methods is to first select the most informative data examples located closer to the class boundary region by using the separating hyperplane found by training an SVM model on the original imbalanced dataset, and then use only those examples in resampling. We demonstrate that it would be possible to obtain comparable classification results to the random oversampling results through two sets of efficient resampling methods which use 50% less amount of data and 75% less amount of data, respectively, compared to the sizes of the datasets generated by the random oversampling method.

Support vector machines (SVMs) is a popular machine learning technique, which works effectively with balanced datasets. However, when it comes to imbalanced datasets, SVMs produce suboptimal classification models. On the other hand, the SVM algorithm is sensitive to outliers and noise present in the datasets. Therefore, although the existing class imbalance learning (CIL) methods can make SVMs less sensitive to class imbalance, they can still suffer from the problem of outliers and noise. Fuzzy SVMs (FSVMs) is a variant of the SVM algorithm, which has been proposed to handle the problem of outliers and noise. In FSVMs, training examples are assigned different fuzzy-membership values based on their importance, and these membership values are incorporated into the SVM learning algorithm to make it less sensitive to outliers and noise. However, like the normal SVM algorithm, FSVMs can also suffer from the problem of class imbalance. In this paper, we present a method to improve FSVMs for CIL (called FSVM-CIL), which can be used to handle the class imbalance problem in the presence of outliers and noise. We thoroughly evaluated the proposed FSVM-CIL method on ten real-world imbalanced datasets and compared its performance with five existing CIL methods, which are available for normal SVM training. Based on the overall results, we can conclude that the proposed FSVM-CIL method is a very effective method for CIL, especially in the presence of outliers and noise in datasets.

In class imbalance learning, the performance measure used for the model selection would play a vital role. It has been well-studied in the past research that the most widely used performance measure, the overall accuracy of the model, can lead to sub-optimal classification models when learning from imbalanced datasets. In order to overcome this problem, other performance measures, such as the geometric-mean (Gm) and F-measure (Fm), have been used for imbalanced dataset learning. Training a classifier system with an imbalanced dataset (where the positive class is the minority class) would usually produce sub-optimal models having a higher specificity (SP) and a lower sensitivity (SE). By applying class imbalance learning methods, we would often be able to increase the SE by sacrificing some amount of SP. In some type of real world imbalanced classification problems, such as the gene finding Bioinformatics problems, it is important to improve the SE as much as possible by keeping the reduction of SP to the minimum. In this paper, we show that with respect to this type of classification problems the existing performance measures used in class imbalance learning (Gm and Fm) can still result in sub-optimal classification models. In order to circumvent these problems, we introduced a new performance measure, called adjusted geometric-mean (AGm). We show, both analytically and empirically on two real-world Bioinformatics datasets, that AGm can perform better than Gm and Fm metrics.

Abstract Motivation: In this article, we show that the classification of human precursor microRNA (pre-miRNAs) hairpins from both genome pseudo hairpins and other non-coding RNAs (ncRNAs) is a common and essential requirement for both comparative and non-comparative computational recognition of human miRNA genes. However, the existing computational methods do not address this issue completely or successfully. Here we present the development of an effective classifier system (named as microPred) for this classification problem by using appropriate machine learning techniques. Our approach includes the introduction of more representative datasets, extraction of new biologically relevant features, feature selection, handling of class imbalance problem in the datasets and extensive classifier performance evaluation via systematic cross-validation methods. Results: Our microPred classifier yielded higher and, especially, much more reliable classification results in terms of both sensitivity (90.02%) and specificity (97.28%) than the exiting pre-miRNA classification methods. When validated with 6095 non-human animal pre-miRNAs and 139 virus pre-miRNAs from miRBase, microPred resulted in 92.71% (5651/6095) and 94.24% (131/139) recognition rates, respectively. Availability: The microPred classifier, the datasets used, and the features extracted are freely available at http://web.comlab.ox.ac.uk/people/ManoharaRukshan.Batuwita/microPred.htm. Contact: [email protected]; [email protected] Supplementary information: Supplementary data are available at Bioinformatics online.

micoRNA (miRNA) is a vital class of non-coding RNA genes, which participates in post-transcriptional gene regulation in eukaryotic cell. Interestingly, some close relationships between miRNA expression levels and several human diseases like cancers have been recently uncovered. Difficulties of identifying miRNAs via direct experimental method due to their special and temporal expression patterns make the computational prediction methods paramount important. Specially, non-comparative computational methods would have the advantage of recognizing species-specific miRNAs that can be missed by comparative methods. In this paper we present a systematic development of an improved classifier system for non-comparative human miRNA gene recognition using effective machine learning techniques.

This paper applies a neural networks (NN) multiobjective learning algorithm called the Minimum Gradient Method (MGM) to filter noise in regression problems. This method is based on the concept that the learning is a bi-objective problem aiming at minimizing the empirical risk (training error) and the function complexity. The complexity is modeled as the norm of the network output gradient. After training, the NN behaves as an adaptive filter which minimizes the cross-validation error. The NN trained with this method can be used to pre-process the data and help reduce the signal-to-noise ratio (SNR). Some results are presented and they show the effectiveness of the proposed approach.

This paper presents a novel approach for dealing with the structural risk minimization (SRM) applied to a general setting of the machine learning problem. The formulation is based on the fundamental concept that supervised learning is a bi-objective optimization problem in which two conflicting objectives should be minimized. The objectives are related to the empirical training error and the machine complexity. In this paper, one general Q-norm method to compute the machine complexity is presented, and, as a particular practical case, the minimum gradient method (MGM) is derived relying on the definition of the fat-shattering dimension. A practical mechanism for parallel layer perceptron (PLP) network training, involving only quasi-convex functions, is generated using the aforementioned definitions. Experimental results on 15 different benchmarks are presented, which show the potential of the proposed ideas.

In this paper, we employ fuzzy techniques for analysing cancer microarray gene expression data. A fuzzy-based ensemble model and a comprehensive fuzzy-based framework for cancer microarray data analysis are being proposed. Our methods were tested on three benchmark microarray cancer data sets, namely Leukemia Cancer Data Set, Colon Cancer Data Set and Lymphoma Cancer Data Set. Comparing to other traditional statistical and machine learning models, the new approach can efficiently tackle several key problems in cancer microarray gene expression data analysis, including highly correlated genes, high dimensionality, highly noisy data and further explanation of the diagnosis results.

All learning algorithms attempt to improve the accuracy of a classification system. However, the effectiveness of such a system is dependent on the heuristic used by the learning paradigm to measure performance. This paper demonstrates that the use of Precision (P) for performance evaluation of imbalanced data sets could lead the solution towards sub-optimal answers. We move onto present a novel performance heuristic, the 'Optimized Precision (OP)', to negate these detrimental effects. We also analyze the impact of these observations on the training performance of ensemble learners and Multi-Classifier Systems (MCS), and provide guidelines for the proper training of multi-classifier systems.

Although several highly accurate blind source separation algorithms have already been proposed in the literature, these algorithms must store and process the whole data set which may be tremendous in some situations. This makes the blind source separation infeasible and not realisable on VLSI level, due to a large memory requirement and costly computation. This paper concerns the algorithms for solving the problem of tremendous data sets and high computational complexity, so that the algorithms could be run on-line and implementable on VLSI level with acceptable accuracy. Our approach is to partition the observed signals into several parts and to extract the partitioned observations with a simple activation function performing only the "shift-and-add" micro-operation. No division, multiplication and exponential operations are needed. Moreover, obtaining an optimal initial de-mixing weight matrix for speeding up the separating time will be also presented. The proposed algorithm is tested on some benchmarks available online. The experimental results show that our solution provides comparable efficiency with other approaches, but lower space and time complexity.

A unified approach for integrating explicit and implicit knowledge in connectionist knowledge-based systems is proposed. The explicit knowledge is represented by discrete fuzzy rules which are directly mapped into an equivalent multi-purpose neural network based on a MAPI neuron. Some methods based upon interactive fuzzy operators are presented in order to extract fuzzy rules from trained neural networks. An architecture for a neural knowledge-based system is proposed as a combination of modules based on data learning and fuzzy rules mapping. The combination of explicit and implicit knowledge modules is viewed as an iterative process in knowledge acquisition and refinement.

This paper proposes a method of rule extraction from ordinary backpropagation neural networks, which do not have a structure that facilitates rule extraction. This method is based on interval propagation across the network. The method of rule extraction uses a procedure for inverting a neural network, which is also presented. A common benchmark control problem was used to test the rule extraction and inversion methods.

Neural networks have been criticized for their lack of human comprehensibility, which make them to appear as black box structures to the user. The paper proposes a mechanism that compiles a neural network into an equivalent set of fuzzy rules. Genetic algorithms are used to find the right structure of the fuzzy model equivalent with the neural network, and then to find the best shape of the membership functions. In order to reduce the number of fuzzy rules, we look for a hierarchical structure of the fuzzy system, considering the relations between the network inputs.