Object recognition is essential for autonomous cars, and the amalgamation of camera and light detection and ranging (LiDAR) sensor data has emerged as a pivotal method for accurate three-dimensional (3D) object recognition. Contemporary algorithms face challenges with fragmented data, high processing costs, insufficient resolution, and limited dynamic information. This study presents a novel approach utilising 3D U-Net deep learning for precise 3D object detection and localisation by integrating camera and LiDAR data. The process involves obtaining and preprocessing camera and LiDAR data, utilising a geometric 3D frustum method to extract 3D information from LiDAR based on 2D camera bounding boxes, and training a You Only Look Once version 4 (YOLO v4) network to recognise these boundaries in camera images. The detected images are combined with LiDAR data, and a deep U-Net network is utilised to define 3D bounding boxes. Performance is assessed at various noise levels (0 %, 1 %, 2 %, 5 %, and 10 %) in the composite images. This method leverages the benefits of both sensors to better object recognition across diverse shapes and sizes, even in challenging situations, signifying a significant progression towards safer and more reliable autonomous vehicles with improved situational awareness in intricate urban environments.

Spinal ultrasound imaging offers a radiation-free and cost-effective alternative to X-rays for scoliosis screening, with advantages such as non-ionizing radiation, affordability, and effective soft tissue visualization. These characteristics make it a promising modality for clinical applications, especially in pediatric and frequent monitoring scenarios. However, inherent challenges—such as low contrast, high speckle noise, and various imaging artifacts—significantly degrade image quality, posing difficulties for downstream tasks like anatomical landmark detection. This limitation hinders the deployment of fully automated AI-based diagnostic systems. AI techniques have been applied to spinal ultrasound imaging in previous studies, employing models such as Faster R-CNN, YOLOv8, and YOLOv11 for vertebrae detection. However, challenges in robustness and anatomical prior modeling remain. To address these challenges, we propose a novel framework for the accurate detection of spinal bones in spinal ultrasound images, based on an enhanced YOLOv11 architecture. Our method introduces (1) a heatmap-based center-point branch modeling spinal locations as 2D Gaussians and (2) an Anatomical Spatial Interaction Module that fuses spatial and semantic features via coordinate injection and attention mechanisms. Evaluated on a clinical spinal ultrasound dataset, our approach outperforms these prior methods, achieving a 2.4-point gain in [email protected]:0.95. These results demonstrate the potential of our framework as an effective tool for radiation-free, AI-assisted spinal assessment.

Recent developments have established ultrasound imaging as a promising new standard for scoliosis assessment. However, its limited penetration in soft tissue inevitably produces acoustic shadowing and loss of deeper structural information in individual B-mode slices. Critical spine-related features could therefore be absent or inconsistently represented across adjacent slices, and effective feature extraction is further undermined by speckle noise, a major additive object, which further decreases local contrast. Together, these limitations impede the detection of key anatomical landmarks such as thoracic bony features (TBF) and lamellar bone features (LBF), blocking accurate downstream analysis. To overcome these challenges, we present the Contextual-Driven Ultrasound Enhancement Network (CD-Net), a self-supervised framework that fuses information across slices and refines local detail. CD-Net comprises two core modules: (1) the Contextual Cross-Attention Transfer (CCAT) module, which captures and transfers interslice spatial relationships, and (2) the Localized Attention Contrast Enhancement (LoCE) module, which selectively sharpens and enhances feature regions. On a dataset of 309 patients, CD-Net boosts the detection rate from 78.25% to 93.18% and achieves a Structural Similarity Index (SSIM) of 89.2% (σ = 0.045). Enhanced visibility of TBF and LBF across varying image qualities demonstrates CD-Net’s potential to significantly improve the reliability and efficiency of ultrasound-based scoliosis diagnosis in clinical practice.

Scoliosis, a three-dimensional spinal deformity, requires early and accurate detection for effective treatment. While Cobb’s angle measurement from radiographs remains the clinical gold standard, the associated radiation exposure underscores the need for safer alternatives. Ultrasound imaging offers a non-invasive solution; however, it presents significant challenges, including low contrast, high noise, and irregular anatomical structures, which complicate the accurate estimation of the Ultrasound Curve Angle (UCA).Previous studies have attempted to improve segmentation performance on ultrasound images, often relying on limited paired datasets. While these methods can enhance results, they risk overfitting due to the small sample size and lack a broader understanding of inter-sample relationships.To address these limitations, we propose ISC-Swin, a Swin Transformer-based framework integrated with inter-sample contrastive learning for more robust spine feature segmentation in ultrasound images. Our architecture leverages both local and global contextual information through a novel Inter-Sample Contrastive Bank (ISCB), which dynamically extracts multilevel features across diverse samples. By explicitly modeling inter-class and intra-class differences, ISC-Swin improves the detection of subtle spinal features in challenging, noisy environments.Experimental results show that ISC-Swin achieves a 1–5% improvement in both Dice Similarity Coefficient and Intersection over Union (IoU) metrics, surpassing current state-of-the-art models in bone feature detection and enhancing diagnostic precision in ultrasound-based scoliosis assessment.

Infrared small target detection is crucial in military reconnaissance, remote sensing, and so on. However, due to its small size and the high coupling with complex backgrounds, the present methods still face challenges in precise detection. They predominantly focus on target feature learning while neglecting the critical role of background modeling for small target decoupling. To this end, we propose a dual-task collaborative optimization network (DCONet), which decouples the task into background estimation and target segmentation using a multistage iterative optimization strategy. First, considering significant directional distribution characteristics in infrared backgrounds, we propose a direction-aware background estimation module (DBEM) to capture directional features, such as clouds and trees, thereby generating an initial background estimation. Second, we propose a background suppression gating unit (BSGU), which employs a gating mechanism and a channel-level adjustment factor to dynamically suppress background noise based on the preliminary background estimation, thereby generating the target segmentation result. Finally, the estimated background, target segmentation, and the reconstructed original image based on them are propagated to the next stage for further iterative optimization. The experimental results show that DCONet performs better than existing methods across three public datasets. The source code is available athttps://github.com/tustAilab/DCONet

In medical image analysis, anatomical landmarks usually contain strong prior knowledge of their structural information. In this paper, we propose to promote medical landmark localization by modeling the underlying landmark distribution via normalizing flows. Specifically, we introduce the flow-based landmark distribution prior as a learnable objective function into a regression-based landmark localization framework. Moreover, we employ an integral operation to make the mapping from heatmaps to coordinates differentiable to further enhance heatmap-based localization with the learned distribution prior. Our proposed Normalizing Flow-based Distribution Prior (NFDP) employs a straightforward backbone and non-problem-tailored architecture (i.e., ResNet18), which delivers high-fidelity outputs across three X-ray-based landmark localization datasets. Remarkably, the proposed NFDP can do the job with minimal additional computational burden as the normalizing flows module is detached from the framework on inferencing. As compared to existing techniques, our proposed NFDP provides a superior balance between prediction accuracy and inference speed, making it a highly efficient and effective approach. The source code of this paper is available at https://github.com/jacksonhzx95/NFDP.

A growing body of literature has proved that general type-2 fuzzy control systems (GT2 FCSs) are reliable, robust, and safe control systems against severe external disturbances, high levels of noise, and uncertainty that are inevitable in real-world applications. Further studies on the GT2 FCSs are therefore needed to provide new insights into these control systems, in particular over their stability and computational complexity problems. Unlike the existing works on stability analysis of GT2 FCSs in the time domain, the aim of this article is to assess the stability properties of these systems in the frequency domain through a novel intuitive method. Before delving into stability analysis, an initial step involves reducing computational complexity. This is achieved by introducing a streamlined version of the GT2 fuzzy controller (GT2 FC). This simplified architecture is constructed using a series of zSlices-based interval type-2 fuzzy-logic controls (zIT2 FLCs) situated at specific zLevels. Each zIT2 FLC is composed of two embedded type-1 fuzzy FLCs (T1 FLCs). The subsequent phase entails a methodical procedure for assessing stability based on the existence of limit cycles. The initial stage involves the linearization of the simplified GT2 FC through the derivation of its describing function (DF). Next, a combination of the parameter plane approach and the particle swarm optimization (PSO) technique is leveraged. These techniques serve the purpose of pinpointing the regions corresponding to limit cycles and asymptotic stability with a focus on improving the stability boundary. Following this, the analysis shifts toward quantifying the system’s resilience in the face of uncertainty. Stability margins required to generate a limit cycle are calculated using stability equations. This step provides a measure of how robust the closed-loop system remains under varying degrees of uncertainty. Finally, three simulation examples are presented to justify the advantages of the proposed approach.

Ultrasound volume projection imaging has shown great promise to visualize spine features and diagnose scoliosis thanks to its harmlessness, cheapness, and efficiency. The key to measuring spine deformity and assessing scoliosis is to accurately segment the spine bone features. In this paper, we propose a novel structure-affinity dual attention-based network (SADANet) for effective spine segmentation. Global channel attention module and spatial criss-cross attention module are combined in a parallel manner to generate rich global context of spine images. Meanwhile, we present a structure-affinity strategy to encode the structural knowledge of spine bones into the semantic representations. By this means, the network can capture both contextual and structural information. Experiments show that our proposed algorithm achieves promising performance on spine segmentation as compared with other state-of-the-art candidates, which makes it an appealing approach for intelligent scoliosis assessment.

Abstract In recent years, there has been a renewal of interest in brain–computer interface (BCI). One of the BCI tasks is to classify the EEG motor imagery (MI). A great deal of effort has been made on MI classification. What seems to be lacking, however, is multiple MI classification. This paper develops a single-channel-based convolutional neural network to tackle multi-classification motor imagery tasks. For multi-classification, a single-channel learning strategy can extract effective information from each independent channel, making the information between adjacent channels not affect each other. A data evaluation method and a mutual information-based regularization parameters auto-selection algorithm are also proposed to generate effective spatial filters. The proposed method can be used to tackle the problem of an inaccurate mixed covariance matrix caused by fixed regularization parameters and invalid training data. To illustrate the merits of the proposed methods, we used the tenfold cross-validation accuracy and kappa as the evaluation measures to test two data sets. BCI4-2a and BCI3a data sets have four mental classes. For the BCI4-2a data set, the average accuracy is 79.01%, and the kappa is 0.7202 using data evaluation-based auto-selected filter bank regularized common spatial pattern voting (D-ACSP-V) and single-channel series convolutional neural network (SCS-CNN). Compared to traditional FBRCSP, the proposed method improved accuracy by 7.14% for the BCI4-2a data set. By using the BCI3a data set, the proposed method improved accuracy by 9.54% compared with traditional FBRCSP, the average accuracy of the proposed method is 83.70%, and the kappa is 0.7827.

Molecular communication (MC) is an emerging communication paradigm that allows bio-nanomachines (NMs) to communicate using biochemical molecules as information carriers. It can be used in many promising biomedical applications such as the Internet of Bio-Nano Things (IoBNT) for targeted drug delivery and healthcare applications. In particular, the blood-tissue barrier (BTB) inside the body forms the main communication pathway for molecular information exchange between the NMs as well as between the intrabody nanonetwork and the bio-cyber interface in the IoBNT network. However, overcoming this barrier by the molecules is one of the main challenges for MC in the body. Therefore, spatiotemporal modeling of MC across the BTB is of particular interest. In this article, we develop a mathematical model and stochastic particle-based simulator for MC over high spatiotemporal resolution between mobile NMs in the blood capillary and the surrounding tissue. The transmitting bio-NM is modeled as a moving sphere with a continuous emission pattern over a specific duration. In this work, the blood capillary characteristics, including the BTB and blood flow, are modeled and their effect is examined on the molecular received signal. In addition, we examined the impact of the emission duration, the elimination rate, and the separation distance on the molecular received signal. The numerical results are verified using the developed particle-based simulator. This work can help in the optimum design and development of the IoBNT systems based on MC for biomedical applications, such as smart drug delivery and health monitoring systems.

Volume Projection Imaging from ultrasound data is a promising technique to visualize spine features and diagnose Adolescent Idiopathic Scoliosis. In this paper, we present a novel multi-task framework to reduce the scan noise in volume projection images and to segment different spine features simultaneously, which provides an appealing alternative for intelligent scoliosis assessment in clinical applications. Our proposed framework consists of two streams: i) A noise removal stream based on generative adversarial networks, which aims to achieve effective scan noise removal in a weakly-supervised manner, i.e., without paired noisy-clean samples for learning; ii) A spine segmentation stream, which aims to predict accurate bone masks. To establish the interaction between these two tasks, we propose a selective feature-sharing strategy to transfer only the beneficial features, while filtering out the useless or harmful information. We evaluate our proposed framework on both scan noise removal and spine segmentation tasks. The experimental results demonstrate that our proposed method achieves promising performance on both tasks, which provides an appealing approach to facilitating clinical diagnosis.

Automatic spine segmentation, based on ultrasound volume projection imaging (VPI), is of great value in clinical applications to diagnose scoliosis in teenagers. In this paper, we propose a novel framework to improve the segmentation accuracy on spine images via structure-enhanced attentive learning. Since the spine bones contain strong prior knowledge of their shapes and positions in ultrasound VPI images, we propose to encode this information into the semantic representations in an attentive manner. We first revisit the self-attention mechanism in representation learning, and then present a strategy to introduce the structural knowledge into the key representation in self-attention. By this means, the network explores both the contextual and structural information in the learned features, and consequently improves the segmentation accuracy. We conduct various experiments to demonstrate that our proposed method achieves promising performance on spine image segmentation, which shows great potential in clinical diagnosis.

Lung cancer is the leading cause of cancer death worldwide. The critical reason for the deaths is delayed diagnosis and poor prognosis. With the accelerated development of deep learning techniques, it has been successfully applied extensively in many real-world applications, including health sectors such as medical image interpretation and disease diagnosis. By combining more modalities that being engaged in the processing of information, multimodal learning can extract better features and improve the predictive ability. The conventional methods for lung cancer survival analysis normally utilize clinical data and only provide a statistical probability. To improve the survival prediction accuracy and help prognostic decision-making in clinical practice for medical experts, we for the first time propose a multimodal deep learning framework for non-small cell lung cancer (NSCLC) survival analysis, named DeepMMSA. This framework leverages CT images in combination with clinical data, enabling the abundant information held within medical images to be associate with lung cancer survival information. We validate our model on the data of 422 NSCLC patients from The Cancer Imaging Archive (TCIA). Experimental results support our hypothesis that there is an underlying relationship between prognostic information and radiomic images. Besides, quantitative results show that our method could surpass the state-of-the-art methods by 4% on concordance.

This article investigates deep neural networks (DNNs)-based lung nodule classification with hyperparameter optimization. Hyperparameter optimization in DNNs is a computationally expensive problem, and a surrogate-assisted evolutionary algorithm has been recently introduced to automatically search for optimal hyperparameter configurations of DNNs, by applying computationally efficient surrogate models to approximate the validation error function of hyperparameter configurations. Different from existing surrogate models adopting stationary covariance functions (kernels) to measure the difference between hyperparameter points, this article proposes a nonstationary kernel that allows the surrogate model to adapt to functions whose smoothness varies with the spatial location of inputs. A multilevel convolutional neural network (ML-CNN) is built for lung nodule classification, and the hyperparameter configuration is optimized by the proposed nonstationary kernel-based Gaussian surrogate model. Our algorithm searches with a surrogate for optimal setting via a hyperparameter importance-based evolutionary strategy, and the experiments demonstrate our algorithm outperforms manual tuning and several well-established hyperparameter optimization methods, including random search, grid search, the tree-structured parzen estimator (TPE) approach, Gaussian processes (GP) with stationary kernels, and the recently proposed hyperparameter optimization via RBF and dynamic (HORD) coordinate search.

3D ultrasound imaging shows great promise for scoliosis diagnosis thanks to its low-costing, radiation-free and real-time characteristics. The key to accessing scoliosis by ultrasound imaging is to accurately segment the bone area and measure the scoliosis degree based on the symmetry of the bone features. The ultrasound images tend to contain many speckles and regular occlusion noise which is difficult, tedious and time-consuming for experts to find out the bony feature. In this paper, we propose a robust bone feature segmentation method based on the U-net structure for ultrasound spine Volume Projection Imaging (VPI) images. The proposed segmentation method introduces a total variance loss to reduce the sensitivity of the model to small-scale and regular occlusion noise. The proposed approach improves 2.3% of Dice score and 1% of AUC score as compared with the u-net model and shows high robustness to speckle and regular occlusion noise.

Gravitational search algorithm (GSA) is a novel technique as compared to other heuristic methods and depends pon the gravitational forces between masses. It showed better performance in terms of convergence but has slow exploitation ability due to the fitness function effect on masses; they are getting heavier after every iteration. Therefore, masses are getting closer to each other and nullify the gravitational forces on each other avoiding them from swiftly exploiting the optimum. In order to solve this problem in this paper, an advanced gravitational search algorithm (AGSA) with modified exploitation strategy is proposed. The reason for the modification is that the agents will reach the optimum point swiftly and the convergence is much faster as compared to the standard and other improved versions of GSA available in the literature. AGSA is also compared with the standard and modified Particle Swarm optimization algorithm in this paper. Five benchmark functions have been implemented to assess the efficiency of the presented algorithm. In addition, a standard, constrained, design problem of a pressure vessel design is also used to examine the efficiency of the proposed technique. Simulation results empirically validated that the presented algorithm has remarkably better results in accordance with convergence and solution stability when compared to the other methods.

In this paper, a combination of Advanced Particle Swarm Optimization (APSO) and Neural Network are presented to compensate the drawbacks of both the techniques and utilize the strong attributes to form a hybrid system called Hybrid Advance Particle Swarm Optimization-Neural Network System (HAPSONNS). APSO is used for the training of the neural network. In the initial phases of the search, PSO has swift convergence for global optimum, but later it suffers from slow convergence around the global optimum position. On the contrary, the gradient method attains prior to convergence around the global optimum point, therefore, attaining better accuracy in terms of convergence. This paper elucidates the usage of APSO applied to feedforward neural network to improve the classification accuracy of the network and also decreases the network training time.

In this paper, advanced particle swarm optimization Algorithm (APSO) with improved velocity updated strategy is presented. The algorithm incorporates an improved velocity update equation so that the particles will reach the optimum point quickly and convergence is much faster than the standard PSO (SPSO) and other improved PSOs in the literature. Five benchmark functions have been selected to evaluate the efficiency of the proposed algorithm. The simulation results demonstrate that the proposed technique has remarkably improved in terms of convergence and solution quality.

This paper presents a two-class electroencephal-ography-based classification for classifying of driver fatigue (fatigue state versus alert state) from 43 healthy participants. The system uses independent component by entropy rate bound minimization analysis (ERBM-ICA) for the source separation, autoregressive (AR) modeling for the features extraction, and Bayesian neural network for the classification algorithm. The classification results demonstrate a sensitivity of 89.7%, a specificity of 86.8%, and an accuracy of 88.2%. The combination of ERBM-ICA (source separator), AR (feature extractor), and Bayesian neural network (classifier) provides the best outcome with a p-value < 0.05 with the highest value of area under the receiver operating curve (AUC-ROC = 0.93) against other methods such as power spectral density as feature extractor (AUC-ROC = 0.81). The results of this study suggest the method could be utilized effectively for a countermeasure device for driver fatigue identification and other adverse event applications.

In this paper, a deep belief network (DBN) has been adopted as an efficient technique to diagnosis the Parkinson's disease (PD). This diagnosis has been established based on the speech signal of the patients. Through the distinguishing and analyzing of the speech signal, the DBN has the ability to diagnose Parkinson's disease. To realize the diagnosis of Parkinson's disease by using DBN, the proposed system has been trained and tested with voices from a number of patients and healthy people. A feature extraction process has been prepared to be inputted to the deep belief network (DBN) which is used to create a template matching of the voices. In this paper, DBN is used to classify the Parkinson's disease which composes two stacked Restricted Boltzmann Machines (RBMs) and one output layer. Two stages of learning need to be applied to optimize the networks' parameters. The first stage is unsupervised learning which uses RBMs to overcome the problem that can cause because of the random value of the initial weights. Secondly, backpropagation algorithm is used as a supervised learning for the fine tuning. To illustrate the effectiveness of the proposed system, the experimental results are compared with different approaches and related works. The overall testing accuracy of the proposed system is 94% which is better than all of the compared methods. In short, the DBN is an effective method to diagnosis Parkinson's disease by using the speech signal.

This paper presents a systematic method to select optimal electroencephalography (EEG) channels for three mental tasks-based brain-computer interface (BCI) classification. A blind source separation (BSS) technique based on independent component analysis (ICA) with its back-projecting of the scalp map was used for selecting the optimal EEG channels. The three mental tasks included: mental letter composing, mental arithmetic and mental Rubik's cube rolling. Based on a power spectral density (PSD), the features of the two-channel EEG data were extracted, and then were classified by Bayesian neural network. The results of the ICA decomposition with the back-projected scalp map showed that the prominent channels could be selected for dominant features from original six EEG channels (C3, C4, P3, P4, O1, O2) to four dominant channels (P3, O1, C4, O2) with the best two EEG channels selection at O1&C4. Two channel combinations classification yielded to the best two EEG channels of O1&C4 with an accuracy of 76.4%, followed by P3&O2 with an accuracy of 74.5%; P3&C4 with an accuracy of 71.9% and O1&O2 with an accuracy of 70%.

Multi-clustering has been widely used. It acts as a pre-training process for identifying protein-ligand binding in structure-based drug design. Then, the Support Vector Machine (SVM) is employed to classify the sites most likely for binding ligands. Three types of attributes are used, namely geometry-based, energy-based, and sequence conservation. Comparison is made on 198 drug-target protein complexes with LIGSITECSC, SURFNET, Fpocket, Q-SiteFinder, ConCavity, and MetaPocket. The results show an improved success rate of up to 86%.

Fuzzy regression methods have commonly been used to develop consumer preferences models, which correlate the engineering characteristics with consumer preferences regarding a new product; the consumer preference models provide a platform, whereby product developers can decide the engineering characteristics in order to satisfy consumer preferences prior to developing the products. Recent research shows that these fuzzy regression methods are commonly used to model customer preferences. However, these approaches have a common limitation in that they do not investigate the appropriate polynomial structure, which includes significant regressors with only significant engineering characteristics; also, they cannot generate interaction or high-order regressors in the models. The inclusion of insignificant regressors is not an effective approach when developing the models. Exclusion of significant regressors may affect the generalization capability of the consumer preference models. In this paper, a novel fuzzy modeling method is proposed, namely fuzzy stepwise regression (F-SR), in order to develop a customer preference model which is structured with an appropriate polynomial, which includes only significant regressors. Based on the appropriate polynomial structure, the fuzzy coefficients are determined using the fuzzy least-squares regression. The developed fuzzy regression model attempts to obtain a better generalization capability using a smaller number of regressors. The effectiveness of the F-SR is evaluated based on two design problems, namely a tea maker design and a solder paste dispenser design. Results show that better generalization capabilities can be obtained compared with the fuzzy regression methods commonly used for new product development. In addition, smaller scale consumer preference models with fewer engineering characteristics can be obtained. Hence, a simpler and more effective product development platform can be provided.

Insulin-dependent diabetes mellitus is classified as Type 1 diabetes and it can be further classified as immune-mediated or idiopathic. Through the analysis of electrocar-diographic (ECG) signals of 15 children with T1DM, an effective hypoglycemia detection system, hybrid rough set based neural network (RNN) is developed by the use of physiological parameters of ECG signal. In order to detect the status of hypoglycemia, the feature of ECG of type 1 diabetics are extracted and classified according to corresponding glucose levels. In this technique, the applied physiological inputs are partitioned into predicted (certain) or random (uncertain) parts using defined lower and boundary of rough regions. In this way, the neural network is designed to deal only with the boundary region which mainly consists of a random part of applied input signal causing inaccurate modeling of the data set. A global training algorithm, hybrid particle swarm optimization with wavelet mutation (HPSOWM) is introduced for parameter optimization of proposed RNN. The experiment is carried out using real data collected at Department of Health, Government of Western Australia. It indicated that the proposed hybrid architecture is efficient for hypoglycemia detection by achieving better sensitivity and specificity with less number of design parameters.

In this paper, we investigate the stability of polynomial fuzzy-model-based (PFMB) control systems, aiming to relax stability conditions by considering the information of membership functions. To facilitate the stability analysis, we propose a general form of approximated membership functions, which is implemented by Taylor series expansion. Taylor series membership functions (TSMF) can be brought into stability conditions such that the relation between membership grades and system states is expressed. To further reduce the con-servativeness, different types of information are taken into account: the boundary of membership functions, the property of membership functions, and the boundary of operating domain. Stability conditions are obtained from Lyapunov stability theory by sum of squares (SOS) approach. Simulation examples demonstrate the effect of each piece of information.

Large imbalanced datasets have introduced difficulties to classification problems. They cause a high error rate of the minority class samples and a long training time of the classification model. Therefore, re-sampling and data size reduction have become important steps to pre-process the data. In this paper, a sampling strategy over a large imbalanced dataset is proposed, in which the samples of the larger class are selected based on fuzzy logic. To further reduce the data size, the evolutionary computational method of CHC is employed. The evaluation is done by applying a Support Vector Machine (SVM) to train a classification model from the re-sampled training sets. From experimental results, it can be seen that our proposed method improves both the F-measure and AUC. The complexity of the classification model is also compared. It is found that our proposed method is superior to all other compared methods.

In this paper, a self-provisioning of communication network service based on a fuzzy particle swarm optimization is proposed to minimize the configuration cost of four layer communication network. An swarm optimization called fuzzy particle swarm optimization (FPSO) is introduced. In this FPSO, the inertia weight of PSO is adaptively determined by a set of fuzzy rule. Also, a cross-mutated operation is presented to drive the solution to escape from local optima where the control parameter of this operation is also governed by a set of fuzzy rule. A performance comparison is given to show the performance of the proposed FPSO on the self-provisioning of communication network service and found that the performance of FPSO is significantly better than that of the existing hybrid PSO methods in a statistical sense.

This paper focuses on the hybridization technology using rough sets concepts and neural computing for decision and classification purposes. Based on the rough set properties, the lower region and boundary region are defined to partition the input signal to a consistent (predictable) part and an inconsistent (random) part. In this way, the neural network is designed to deal only with the boundary region, which mainly consists of an inconsistent part of applied input signal causing inaccurate modeling of the data set. Owing to different characteristics of neural network (NN) applications, the same structure of conventional NN might not give the optimal solution. Based on the knowledge of application in this paper, a block-based neural network (BBNN) is selected as a suitable classifier due to its ability to evolve internal structures and adaptability in dynamic environments. This architecture will systematically incorporate the characteristics of application to the structure of hybrid rough-block-based neural network (R-BBNN). A global training algorithm, hybrid particle swarm optimization with wavelet mutation is introduced for parameter optimization of proposed R-BBNN. The performance of the proposed R-BBNN algorithm was evaluated by an application to the field of medical diagnosis using real hypoglycemia episodes in patients with Type 1 diabetes mellitus. The performance of the proposed hybrid system has been compared with some of the existing neural networks. The comparison results indicated that the proposed method has improved classification performance and results in early convergence of the network.

This paper presents the classification of a three-class mental task-based brain-computer interface (BCI) that uses the Hilbert-Huang transform for the features extractor and fuzzy particle swarm optimization with cross-mutated-based artificial neural network (FPSOCM-ANN) for the classifier. The experiments were conducted on five able-bodied subjects and five patients with tetraplegia using electroencephalography signals from six channels, and different time-windows of data were examined to find the highest accuracy. For practical purposes, the best two channel combinations were chosen and presented. The three relevant mental tasks used for the BCI were letter composing, arithmetic, and Rubik's cube rolling forward, and these are associated with three wheelchair commands: left, right, and forward, respectively. An additional eyes closed task was collected for testing and used for on-off commands. The results show a dominant alpha wave during eyes closure with average classification accuracy above 90%. The accuracies for patients with tetraplegia were lower compared to the able-bodied subjects; however, this was improved by increasing the duration of the time-windows. The FPSOCM-ANN provides improved accuracies compared to genetic algorithm-based artificial neural network (GA-ANN) for three mental tasks-based BCI classifications with the best classification accuracy achieved for a 7-s time-window: 84.4% (FPSOCM-ANN) compared to 77.4% (GA-ANN). More comparisons on feature extractors and classifiers were included. For two-channel classification, the best two channels were O1 and C4, followed by second best at P3 and O2, and third best at C3 and O2. Mental arithmetic was the most correctly classified task, followed by mental Rubik's cube rolling forward and mental letter composing.

Imbalanced datasets are commonly encountered in real-world classification problems. However, many machine learning algorithms are originally designed for well-balanced datasets. Re-sampling has become an important step to preprocess imbalanced dataset. It aims at balancing the datasets by increasing the sample size of the smaller class or decreasing the sample size of the larger class, which are known as over-sampling and under-sampling respectively. In this paper, a novel sampling strategy based on both over-sampling and under-sampling is proposed, in which the new samples of the smaller class are created by the Synthetic Minority Over-sampling Technique (SMOTE). The improvement of the datasets is done by the evolutionary computational method of CHC that works on both the minority class and majority class samples. The result is a hybrid data preprocessing method that combines both over-sampling and under-sampling techniques to re-sample datasets. The evaluation is done by applying the learning algorithm C4.5 to obtain a classification model from the re-sampled datasets. Experimental results reported that the proposed approach can decrease the over-sampling rate about 50% with only around 3% discrepancy on the accuracy.

Identification of protein-ligand binding site is an important task in structure-based drug design and docking algorithms. In the past two decades, different approaches have been developed to predict the binding site, such as the geometric, energetic, and sequence-based methods. When scores are calculated from these methods, the algorithm for doing classification becomes very important and can affect the prediction results greatly. In this paper, the support vector machine (SVM) is used to cluster the pockets that are most likely to bind ligands with the attributes of geometric characteristics, interaction potential, offset from protein, conservation score, and properties surrounding the pockets. Our approach is compared to LIGSITE, LIGSITE(CSC), SURFNET, Fpocket, PocketFinder, Q-SiteFinder, ConCavity, and MetaPocket on the data set LigASite and 198 drug-target protein complexes. The results show that our approach improves the success rate from 60 to 80 percent at AUC measure and from 61 to 66 percent at top 1 prediction. Our method also provides more comprehensive results than the others.

Hypoglycemia (or low blood glucose) is dangerous for Type 1 diabetes mellitus (T1DM) patients, as this can cause unconsciousness or even death. However, it is impossible to monitor the hypoglycemia by measuring patients' blood glucose levels all the time, especially at night. In this paper, a hypoglycemic episode diagnosis system is proposed to determine T1DM patients' blood glucose levels based on these patients' physiological parameters which can be measured online. It can be used not only to diagnose hypoglycemic episodes in T1DM patients, but also to generate a set of rules, which describe the domains of physiological parameters that lead to hypoglycemic episodes. The hypoglycemic episode diagnosis system addresses the limitations of the traditional neural network approaches which cannot generate implicit information. The performance of the proposed hypoglycemic episode diagnosis system is evaluated by using real T1DM patients' data sets collected from the Department of Health, Government of Western Australia, Australia. Results show that satisfactory diagnosis accuracy can be obtained. Also, explicit knowledge can be produced such that the deficiency of traditional neural networks can be overcome. A clear understanding of how they perform diagnosis can be indicated.

This paper presents a novel fuzzy particle swarm optimization with cross-mutated operation (FPSOCM), where a fuzzy logic is applied to determine the inertia weight of PSO and the control parameter of the proposed cross-mutated operation based on human knowledge. By introducing the fuzzy system, the value of the inertia weight of PSO becomes adaptive. The new cross-mutated operation effectively drives the solution to escape from local optima. To illustrate the performance of the FPSOCM, a suite of benchmark test functions are employed. Experimental results show the proposed FPSOCM method performs better than some existing hybrid PSO methods in terms of solution quality and solution reliability (standard deviation upon many trials). Moreover, an industrial application of economic load dispatch is given to show that the FPSOCM method performs statistically more significant than the existing hybrid PSO methods.

In this paper, multiple detection engines with multi-layered intrusion detection mechanisms are proposed for enhancing computer security. The principle is to coordinate the results from each single-engine intrusion alert system, which seamlessly integrates with a multiple layered distributed service-oriented structure. An improved hidden Markov model (HMM) is created for the detection engine which is capable of the immunology-based self/nonself discrimination. The classifications of normal and abnormal behaviours of system calls are further examined by an advanced fuzzy-based inference process tuned by HPSOWM. Considering a real benchmark dataset from the public domain, our experimental results show that the proposed scheme can greatly shorten the training time of HMM and significantly reduce the false positive rate. The proposed HPSOWM works especially well for the efficient classification of unknown behaviors and malicious attacks.

Severe hypoglycemia is potentially life-threatening. This article introduces a novel hypoglycemia detection strategy using a hybrid particle swarm - based fuzzy support vector machine (SFisSvm) technique. The inputs of this system are six electrocardiographic (ECG) parameters. The system parameters of SFisSvm are optimized using a particle swarm optimization method. The proposed hypoglycemia detector system is a combination of two subsystems, namely, fuzzy inference system (FIS) and support vector machine (SVM). Two most significant inputs, heart rate and RTpc are fed to FIS, and its output is used for input of the SVM. The other ECG parameters and the output of FIS are fed to SVM and, then, are classified to indicate the presence of hypoglycemia. In this study, three and five membership functions are investigated for FIS. Furthermore, radial basis function (RBF), sigmoid and linear kernel functions are employed for mapping the inputs to high dimensional space in SVM. Performances of SFisSvm with different kernel functions are compared. As conclusion, the performance of SFisSvm is found with 75.19%, 83.71% and 79.33% in terms of sensitivity, specificity and geometric mean.

A genetic algorithm (GA)-based neural network classification in the application of brain computer interface (BCI) for controlling a wheelchair is presented in this paper. This study uses an electroencephalography (EEG) as a non-invasive BCI approach to discriminate three non-motor imagery mental tasks for disabled individuals who may have difficulty in using BCI based motor imagery tasks. The three tasks classification is mapped into three wheelchair movements: left, right and forward and the relevant combination mental tasks used in this study are mental arithmetic, letter composing, Rubik's cube rolling, visual counting, ringtone imagery and spatial navigation. The results show the proposed system provides good classification performance after selecting the most effective of three discriminative tasks across combination of the different non-motor imagery mental tasks for the five subjects tested. The average classification accuracy is between 76% and 85 %, with information transfer rates varies from 0.5 to 0.8 bits per trial.

For patients with Type 1 Diabetes Mellitus (T1DM), hypoglycemia or the state of low blood glucose level is a very common but dangerous complication. Hypoglycemia episodes can lead to a large number of serious symptoms and effects, including unconsciousness, coma and even death. The variety of hypoglycemia symptoms is originated from the inadequate supply of glucose to the brain. By analyzing electroencephalography (EEG) signals from five T1DM patients during an overnight study, we find that under hypoglycemia, both centroid theta frequency and centroid alpha frequency change significantly against non-hypoglycemia conditions. Furthermore, a neural network is developed to detect hypoglycemia using the mentioned two EEG features. A standard particle swarm optimization strategy is applied to optimize the parameters of this neural network. By using the proposed method, we obtain the classification performance of 82% sensitivity and 63% specificity. The results demonstrate that hypoglycemia episodes can be detected non-invasively and effectively from EEG signals.

In this study, a normalized radial basis function neural network (NRBFNN) is presented for detection of hypoglycemia episodes by using physiological parameters of electrocardiogram (ECG) signal. Hypoglycemia is a common and serious side effect of insulin therapy in patients with Type 1 diabetes. Based on heart rate (HR) and corrected QT interval (QTc) of electrocardiogram (ECG) signal, a hybrid particle swarm optimization based normalized RBFNN is developed for recognization of hypoglycemia episodes. A global learning algorithm called hybrid particle swarm optimization with wavelet mutation (HPSOWM) is used to optimize the parameters of NRBFNN. From a clinical study of 15 children with Type 1 diabetes, natural occurrence of nocturnal hypoglycemic episodes associated with increased heart rates and corrected QT interval are studied. The overall data are organized into a training set (5 patients), validation set (5 patients) and testing set (5 patients) randomly selected. Using the optimized NRBFNN, the testing performance for detection of hypoglycemic episodes are satisfactory with 76.74% of sensitivity and 51.82% of specificity.

Identification of protein-ligand binding site is an important task in structure-based drug design and docking algorithms. In these two decades, many different approaches have been developed to predict the binding site, such as geometric, energetic and sequence-based methods. When the scores are calculated from these methods, the method of classification is very important and can affect the prediction results greatly. A developed support vector machine (SVM) is used to classify the pockets, which are most likely to bind ligands with the attributes of grid value, interaction potential, offset from protein, conservation score and the information around the pockets. Since SVM is sensitive to the input parameters and the positive samples are more relevant than negative samples, differential evolution (DE) is applied to find out the suitable parameters for SVM. We compare our algorithm to four other approaches: LIGSITE, SURFNET, PocketFinder and Concavity. Our algorithm is found to provide the highest success rate.

Hypoglycemia, or low blood glucose, is the most common complication experienced by Type 1 diabetes mellitus (T1DM) patients. It is dangerous and can result in unconsciousness, seizures and even death. The most common physiological parameter to be effected from hypoglycemic reaction are heart rate (HR) and correct QT interval (QTc) of the electrocardiogram (ECG) signal. Based on physiological parameters, a genetic algorithm based fuzzy reasoning model is developed to recognize the presence of hypoglycemia. To optimize the parameters of the fuzzy model in the membership functions and fuzzy rules, a genetic algorithm is used. A validation strategy based adjustable fitness is introduced in order to prevent the phenomenon of overtraining (overfitting). For this study, 15 children with 569 sampling data points with Type 1 diabetes volunteered for an overnight study. The effectiveness of the proposed algorithm is found to be satisfactory by giving better sensitivity and specificity compared with other existing methods for hypoglycemia detection.

For over two decades, speech control mechanisms have been widely applied in manufacturing systems such as factory automation, warehouse automation, and industrial robotic control for over two decades. To implement speech controls, a commercial speech recognizer is used as the interface between users and the automation system. However, users' commands are often contaminated by environmental noise which degrades the performance of speech recognition for controlling automation systems. This paper presents a multichannel signal enhancement methodology to improve the performance of commercial speech recognizers. The proposed methodology aims to optimize speech recognition accuracy of a commercial speech recognizer in a noisy environment based on a beamformer, which is developed by an intelligent particle swarm optimization. It overcomes the limitation of the existing signal enhancement approaches whereby the parameters inside commercial speech recognizers are required to be tuned, which is impossible in a real-world situation. Also, it overcomes the limitation of the existing optimization algorithm including gradient descent methods, genetic algorithms and classical particle swarm optimization that are unlikely to develop optimal beamformers for maximizing speech recognition accuracy. The performance of the proposed methodology was evaluated by developing beamformers for a commercial speech recognizer, which was implemented on warehouse automation. Results indicate a significant improvement regarding speech recognition accuracy.

In this paper, process conditions of epoxy dispensing processes are determined by the proposed genetic algorithm based neural fuzzy networks, which consists of two tasks: a) the approach of neural fuzzy networks, which was shown to be better than the other existing approaches, is proposed to develop models in relating between process parameters and quality characteristics for the epoxy dispensing processes; b) the approach of genetic algorithm is used to determine process parameters with respect to pre-defined quality requirements based on the developed neural fuzzy network models. The results indicate that, based on the proposed genetic algorithm based neural fuzzy network, estimated process parameters can achieve specified requirements of microchip encapsulations with high and robust qualities.

Fuzzy regression is a commonly used approach for modeling manufacturing processes in which the availability of experimental data is limited. Fuzzy regression can address fuzzy nature of experimental data in which fuzziness is not avoidable while carrying experiments. However, fuzzy regression can only address linearity in manufacturing process systems, but nonlinearity, which is unavoidable in the process, cannot be addressed. In this paper, an evolutionary fuzzy regression which integrates the mechanism of a fuzzy regression and genetic programming is proposed to generate manufacturing process models. It intends to overcome the deficiency of the fuzzy regression, which cannot address nonlinearities in manufacturing processes. The evolutionary fuzzy regression uses genetic programming to generate the structural form of the manufacturing process model based on tree representation which can address both linearity and nonlinearities in manufacturing processes. Then it uses a fuzzy regression to determine outliers in experimental data sets. By using experimental data excluding the outliers, the fuzzy regression can determine fuzzy coefficients which indicate the contribution and fuzziness of each term in the structural form of the manufacturing process model. To evaluate the effectiveness of the evolutionary fuzzy regression, a case study regarding modeling of epoxy dispensing process is carried out.

The limited power consumption, as a major constraint, presents challenges in improving the network throughput for Wireless Sensor Networks (WSNs). Due to the limited computational power, the applications of WSNs in Terrestrial Networks require the capability to pre-process the observation data so as to remove irrelevant features or factors from multi-dimensional dataset. This paper proposes a intelligent distributed energy efficient routing algorithm inspired from natural learning and adaptation process with the aid of hybrid Neural Rough Sets theory, which is used to efficiently reduce the dimensionality of input dataset. The algorithmic implementation and experimental validation are described in this paper. Details of the algorithm and its testing procedures are presented in comparison with the other power-aware protocols, e.g., mini-hop. The validation of the proposed model is carried out via a wireless sensor network test-bed implemented in Castalia Simulator. The experimental results show the network performance measurements such as delay, throughput and packet loss that have been greatly improved as the outcome of applying this integration with Neural Rough Sets.

In this paper, Differential Evolution (DE) that incorporates fuzzy control and k-nearest neighbors algorithm is proposed to tackle the economic load dispatch problem. To provide the self-terminating ability, a technique called Iteration Windows (IW) is introduced to govern the number of iteration in each searching stage during the optimization. The size of IW is controlled by a fuzzy controller, which uses the information provided by the k-nearest neighbors system to analyze the population during the searching process. The controller keeps controlling the IW till the end of the searching process. A wavelet based mutation process is embedded in the DE searching process to enhance the searching performance. The weight F of DE is also controlled by the fuzzy controller to further speed up the searching process. The proposed method is employed to solve the Economic Load Dispatch with Valve-Point Loading (ELD-VPL) Problem. It is shown empirically that the proposed method can terminate the searching process with a reasonable number of iteration and performs significantly better than the conventional methods in terms of convergence speed and solution quality.

A fuzzy particle swarm optimization (PSO) for the minimization of makespan in permutation flow shop scheduling problem is presented in this paper. In the proposed fuzzy PSO, the inertia weight of PSO and the control parameter of the cross mutated operation are determined by a set of fuzzy rules. To escape the local optimum, cross-mutated operation is introduced. In order to make PSO suitable for solving permutation flow shop scheduling problem, a roulette wheel mechanism is proposed to convert the continuous position values of particles to job per mutations. Meanwhile, a swap-based local search for scheduling problem is designed for the local exploration on a discrete job permutation space. Flow shop benchmark functions are employed to evaluate the performance of the fuzzy PSO for flow shop scheduling problems and the results indicate that the algorithm performs better compared with existing hybrid PSO algorithms.

In this paper, we develope a genetic algorithm based fuzzy inference system to recognize hypoglycemic episodes based on heart rate and corrected QT interval of the electrocardiogram (ECG) signal. Genetic algorithm is introduced to optimize the membership functions and fuzzy rules. A practical experiment based on data from 15 children with T1DM is studied. All the data sets are collected from the Department of Health, Government of Western Australia. To prevent the phenomenon of overtraining (over-fitting), a validation strategy that may adjust the fitness function is proposed. Thus, the data are organized into a training set, a validation set, and a testing set randomly selected. The classification results in term of sensitivity, specificity, and receiver operating characteristic (ROC) analysis show that the proposed classification method performs well.

This paper, proposes a hybrid fuzzy logic-based particle swarm optimization (PSO) with cross-mutated operation method for the minimization of makespan in permutation flow shop scheduling problem. This problem is a typical non-deterministic polynomial-time (NP) hard combinatorial optimization problem. In the proposed hybrid PSO, fuzzy inference system is applied to determine the inertia weight of PSO and the control parameter of the proposed cross-mutated operation by using human knowledge. By introducing the fuzzy system, the inertia weight becomes adaptive. The cross-mutated operation effectively forces the solution to escape the local optimum. To make PSO suitable for solving flow shop scheduling problem, a sequence-order system based on the roulette wheel mechanism is proposed to convert the continuous position values of particles to job permutations. Meanwhile, a new local search technique namely swap-based local search for scheduling problem is designed and incorporated into the hybrid PSO. Finally, a suite of flow shop benchmark functions are employed to evaluate the performance of the proposed PSO for flow shop scheduling problems. Experimental results show empirically that the proposed method outperforms the existing hybrid PSO methods significantly.

Hypoglycemia or low blood glucose is dangerous and can result in unconsciousness, seizures, and even death. It is a common and serious side effect of insulin therapy in patients with diabetes. Hypoglycemic monitor is a noninvasive monitor that measures some physiological parameters continuously to provide detection of hypoglycemic episodes in type 1 diabetes mellitus patients (T1DM). Based on heart rate (HR), corrected QT interval of the ECG signal, change of HR, and the change of corrected QT interval, we develop a genetic algorithm (GA)-based multiple regression with fuzzy inference system (FIS) to classify the presence of hypoglycemic episodes. GA is used to find the optimal fuzzy rules and membership functions of FIS and the model parameters of regression method. From a clinical study of 16 children with T1DM, natural occurrence of nocturnal hypoglycemic episodes is associated with HRs and corrected QT intervals. The overall data were organized into a training set (eight patients) and a testing set (another eight patients) randomly selected. The results show that the proposed algorithm performs a good sensitivity with an acceptable specificity.

Hypoglycemia is dangerous for Type 1 diabetes mellitus (T1DM) patients. Based on the physiological parameters, we have developed a classification unit with hybridizing the approaches of neural networks and genetic algorithm to identify the presences of hypoglycemic episodes for TIDM patients. The proposed classification unit is built and is validated by using the real T1DM patients' data sets collected from Department of Health, Government of Western Australia. Experimental results show that the proposed neural network based classification unit can achieve more accurate results on both trained and unseen T1DM patients' data sets compared with those developed based on the commonly used classification methods for medical diagnosis including statistical regression, fuzzy regression and genetic programming.

In this paper, a hybrid evolutionary algorithm (HEA) based on the approaches of the evolutionary algorithm and a local search (LS) is proposed to determine the gene signatures for predicting histologic response of chemotherapy on osteosarcoma patients, which is one of the most common malignant bone tumor in children. The HEA consists of a population of individuals but the evolution of individuals is conducted by a LS, rather than the crossover and mutation used in the traditional evolutionary algorithms. The proposed HEA can simultaneously optimize the feature subset and the classifier through a common solution coding mechanism. Experimental results indicate that HEA can obtain more accurate signatures than the other existing approaches in determining chemoresponse for osteosarcoma.

A new Differential Evolution (DE) that incorporates fuzzy control and k-nearest neighbors algorithm to determine the terminating condition is proposed. A technique called Iteration Windows is introduced to govern the number of iteration in each searching stage. The size of the iteration windows is controlled by a fuzzy controller, which uses the information provided by the k-nearest neighbors system to analyze the population during the searching process. The controller keeps controlling the iteration windows until the end of the searching process. The wavelet based mutation process is embedded in the DE searching process to enhance the searching performance of DE. The F weight of DE is also controlled by the fuzzy controller to further speed up the searching process. A suite of benchmark test functions is employed to evaluate the performance of the proposed method. It is shown empirically that the proposed method can terminate the searching process with a reasonable number of iteration.

In this paper, a modified Differential Evolution (DE) that incorporates double wavelet-based operations is proposed to handle a load flow problem. The wavelet based operation is embedded in the DE mutation and crossover operation. In the DE mutation operation, the scaling factor is controlled by a wavelet function. In the DE crossover operation, a wavelet-based mutation operation is embedded in it. The trial population vectors are thus modified by the wavelet function. The double wavelet mutations are applied in order to enhance DE in exploring the high-dimension solution space more effectively for better solution quality and stability. The proposed DE algorithm is employed to solve the Economic Load Dispatch with Valve-Point Loading (ELD-VPL) Problem. It is shown empirically that the proposed method out-performs significantly the conventional methods in terms of convergence speed, solution quality and solution stability.

Low blood glucose (Hypoglycaemia) is dangerous and can result in unconsciousness, seizures and even death. It has a common and serious side effect of insulin therapy in patients with diabetes. We measure physiological parameters (heart rate, corrected QT interval of the electrocardiogram (ECG) signal, change of heart rate, and the change of corrected QT interval) continuously to provide detection of hypoglycaemic. Based on these physiological parameters, we have developed a genetic algorithm based multiple regression model to determine the presence of hypoglycaemic episodes. Genetic algorithm is used to determine the optimal parameters of the multiple regression. The overall data were organized into a training set (8 patients) and a testing set (another 8 patient) which are randomly selected. The clinical results show that the proposed algorithm can achieve predictions with good sensitivities and acceptable specificities.

An improved Differential Evolution (DE) that incorporates a wavelet-based mutation operation to control the scaling factor is proposed. The wavelet theory applied is to enhance DE in exploring the solution spaces more effectively for better solutions. A suite of benchmark test functions is employed to evaluate the performance of the proposed method. It is shown empirically that the proposed method outperforms significantly the existing methods in terms of convergence speed, solution quality and solution stability.

To improve cancer diagnosis and drug development, the classification of tumor types based on genomic information is important. As DNA microarray studies produce a large amount of data, expression data are highly redundant and noisy, and most genes are believed to be uninformative with respect to the studied classes. Only a fraction of genes may present distinct profiles for different classes of samples. Classification tools to deal with these issues are thus important. These tools should learn to robustly identify a subset of informative genes embedded in a large dataset that is contaminated with high dimensional noises. In this paper, an integrated approach of support vector machine (SVM) and particle swarm optimization (PSO) is proposed for this purpose. The proposed approach can simultaneously optimize the selection of feature subset and the classifier through a common solution coding mechanism. As an illustration, the proposed approach is applied to search the combinational gene signatures for predicting histologic response to chemotherapy of osteosarcoma patients. Cross-validation results show that the proposed approach outperforms other existing methods in terms of classification accuracy. Further validation using an independent dataset shows misclassification of only one out of fourteen patient samples, suggesting that the selected gene signatures can reflect the chemoresistance in osteosarcoma.

This paper proposes a genetic algorithm (GA) based beamformer to optimize speech recognition accuracy for a pretrained speech recognizer. The proposed beamformer is designed to tackle the non-differentiable and non-linear natures of speech recognition by employing the GA algorithm to search for the optimal beamformer weights. Specifically, a population of beamformer weights is reproduced by crossover and mutation until the optimal beamformer weights are obtained. Results show that the speech recognition accuracies can be greatly improved even in noisy environments.

This paper presents a new particle swarm optimization (PSO) algorithm for tuning parameters (weights) of neural networks. The new PSO algorithm is called fuzzy logic-based particle swarm optimization with cross-mutated operation (FPSOCM), where the fuzzy inference system is applied to determine the inertia weight of PSO and the control parameter of the proposed cross-mutated operation by using human knowledge. By introducing the fuzzy system, the value of the inertia weight becomes variable. The cross-mutated operation is effectively force the solution to escape the local optimum. Tuning parameters (weights) of neural networks is presented using the FPSOCM. Numerical example of neural network is given to illustrate that the performance of the FPSOCM is good for tuning the parameters (weights) of neural networks.

Classification of tumor types based on genomic information is essential for improving future cancer diagnosis and drug development. Since DNA microarray studies produce a large amount of data, effective analytical methods have to be developed to sort out whether specific cancer samples have distinctive features of gene expression over normal samples or other types of cancer samples. In this paper, an integrated approach of support vector machine (SVM) and genetic algorithm (GA) is proposed for this purpose. The proposed approach can simultaneously optimize the feature subset and the classifier through a common solution coding mechanism. As an illustration, the proposed approach is applied in searching the combinational gene signatures for predicting histologic response to chemotherapy of osteosarcoma patients, which is the most common malignant bone tumor in children. Cross-validation results show that the proposed approach outperforms other existing methods in terms of classification accuracy. Further validation using an independent dataset shows misclassification of only one of fourteen patient samples suggesting that the selected gene signatures can reflect the chemoresistance in osteosarcoma.

An hybrid Particle Swarm Optimization PSO-based wavelet neural network for modelling the development of fluid dispensing for electronic packaging is presented in this paper. In modelling the fluid dispensing process, it is important to understand the process behaviour as well as determine optimum operating conditions of the process for a high-yield, low cost and robust operation. Modelling the fluid dispensing process is a complex non-linear problem. This kind of problem is suitable to be solved by neural network. Among different kinds of neural networks, the wavelet neural network is a good choice to solve the problem. In the proposed wavelet neural network, the translation parameters are variables depending on the network inputs. Thanks to the variable translation parameters, the network becomes an adaptive one. Thus, the proposed network provides better performance and increased learning ability than conventional wavelet neural networks. An improved hybrid PSO [1] is applied to train the parameters of the proposed wavelet neural network. A case study of modelling the fluid dispensing process on electronic packaging is employed to demonstrate the effectiveness of the proposed method.

This paper presents a method on how to estimate main effects of gene representation. This estimate can be used not only to understand the domination of genes in the representation but also to design the mutation rate in genetic algorithms (GAs). A new approach of dynamic mutation rate is proposed by integrating the information of the main effects into the genes. By introducing the proposed method in GAs, both solution quality and solution stability can be improved in solving a set of parametrical test functions. The algorithm was applied to two illustrative applications to evaluate the performance of the proposed method, where the first application is on solving uncapacitated facility location problems and the next is on optimal power flow problems, which are employed. Results indicate that the proposed method yields significantly better results than the existing methods.

This paper presents a neural-tuned neural network (NTNN), which is trained by an improved genetic algorithm (GA). The NTNN consists of a common neural network and a modified neural network (MNN). In the MNN, a neuron model with two activation functions is introduced. An improved GA is proposed to train the parameters of the proposed network. A set of improved genetic operations are presented, which show superior performance over the traditional GA. The proposed network structure can increase the search space of the network and offer better performance than the traditional feed-forward neural network. Two application examples are given to illustrate the merits of the proposed network and the improved GA.

A new hybrid particle swarm optimization (PSO) that incorporates a wavelet-theory-based mutation operation is proposed. It applies the wavelet theory to enhance the PSO in exploring the solution space more effectively for a better solution. A suite of benchmark test functions and three industrial applications (solving the load flow problems, modeling the development of fluid dispensing for electronic packaging, and designing a neural-network-based controller) are employed to evaluate the performance and the applicability of the proposed method. Experimental results empirically show that the proposed method significantly outperforms the existing methods in terms of convergence speed, solution quality, and solution stability.

In this paper, an improved genetic algorithm has been proposed for solving multi-contingency transient stability constrained optimal power flow (MC-TSCOPF) problems. The MC-TSCOPF problem is formulated as an extended optimal power flow (OPF) with additional generator rotor angle constraints and is converted into an unconstrained optimization problem, which is suitable for genetic algorithms to deal with, using a penalty function. The improved genetic algorithm is proposed by incorporating an orthogonal design in exploring solution spaces. A case study indicates that the improved genetic algorithm outperforms the existing genetic algorithm-based method in terms of robustness of solutions and the convergence speed while the solution quality can be kept.

Determination of the initial process parameters for fluid dispensing process is a highly skilled task and is usually based on skilled engineers’ intuitive sense acquired through long-term experience rather than on a knowledge-based approach. In the face of global competition, the current trial-and -error practice is inadequate. In this paper, a rule-based system is developed to aid the determination of initial process parameters for fluid dispensing process by the genetic algorithm. Based on the rule based system, a set of ranges of process parameters can be recommended with a pre-defined quality requirement of microchip encapsulation. The preliminary validation test of the rule-based system has indicated that it can determine a set of ranges of initial process parameters for fluid dispensing process effectively, from which quality requirement can be achieved without totally relying on engineers’ experience.

This paper presents the control of nonlinear systems with a neural network. In the proposed neural network, the neuron has two activation functions and exhibits a node-to-node relationship in the hidden layer. By using a genetic algorithm with arithmetic crossover and non-uniform mutation, the parameters of the proposed neural network can be tuned. Application examples are given to illustrate the merits of the proposed neural network.

An improved hybrid particle swarm optimization (PSO) that incorporates a wavelet-based mutation operation is proposed It applies wavelet theory to enhance PSO in exploring solution spaces more effectively for better solutions. A suite of benchmark test functions and an application example on tuning an associative-memory neural network are employed to evaluate the performance of the proposed method. It is shown empirically that the proposed method outperforms significantly the existing methods in terms of convergence speed, solution quality and solution stability.

This paper presents a variable node-to-node-link neural network (VN2NN) trained by real-coded genetic algorithm (RCGA). The VN2NN exhibits a node-to-node relationship in the hidden layer, and the network parameters are variable. These characteristics make the network adapt to the changes of the input environment, enable it to tackle different input sets distributed in a large domain. Each input data set is effectively handled by a corresponding set of network parameters. The set of parameters are governed by the other nodes. Taking the advantage of these features, the proposed network ensures better learning and generalization abilities. Application of the proposed network to hand-written graffiti recognition will be presented so as to illustrate the improvement.

This paper presents the learning of neural network parameters using a real-coded genetic algorithm (RCGA) with proposed crossover and mutation. They are called the average-bound crossover (AveBXover) and wavelet mutation (WM). By introducing the proposed genetic operations, both the solution quality and stability are better than the RCGA with conventional genetic operations. A suite of benchmark test functions are used to evaluate the performance of the proposed algorithm. An application example on an associative memory neural network is used to show the learning performance brought by the proposed RCGA.

A variable translation wavelet neural network (VTWNN) trained by genetic algorithm is presented in this paper. In the proposed wavelet neural network, the translation parameters are variables depending on the network inputs. Thanks to the variable translation parameter, the network becomes an adaptive one, providing better performance and increased learning ability than conventional wavelet neural networks. Genetic algorithm is applied to train the parameters of the proposed wavelet neural network. An application example on short-term daily electric load forecasting in Hong Kong is presented to show the merits of the proposed network.

This paper presents a neural network with variable parameters. These variable parameters adapt to the changes of the input environment, and tackle different input data sets in a large domain. Each input data set is effectively handled by its corresponding set of network parameters. Thus, the proposed neural network exhibits a better learning and generalization ability than a traditional one. An improved genetic algorithm (Lam et al., 2004) is proposed to train the network parameters. An application example on hand-written pattern recognition will be presented to verify and illustrate the improvement.

The paper presents an electric load balancing system for domestic use. An electric load forecasting system, which is realized by a genetic algorithm-based modified neural network, is employed. On forecasting the home power consumption profile, the load balancing system can adjust the amount of energy stored in battery accordingly, preventing it from reaching certain practical limits. A steady consumption from the AC mains can then be obtained which will benefit both the users and the utility company. An example will be given to illustrate the merits of the forecaster, and its performance on achieving the load balancing.

This paper presents a fuzzy-tuned neural network, which is trained by an improved genetic algorithm (GA). The fuzzy-tuned neural network consists of a neural-fuzzy network and a modified neural network. In the modified neural network, a neuron model with two activation functions is used so that the degree of freedom of the network function can be increased. The neural-fuzzy network governs some of the parameters of the neuron model. It will be shown that the performance of the proposed fuzzy-tuned neural network is better than that of the traditional neural network with a similar number of parameters. An improved GA is proposed to train the parameters of the proposed network. Sets of improved genetic operations are presented. The performance of the improved GA will be shown to be better than that of the traditional GA. Some application examples are given to illustrate the merits of the proposed neural network and the improved GA.

This paper presents the estimation of the transmission gain for an AC power line data network in an intelligent home. The estimated gain ensures the transmission reliability and efficiency. A neural-fuzzy network with rule switches is proposed to perform the estimation. An improved genetic algorithm is proposed to tune the parameters and the rules of the proposed neural-fuzzy network. By turning on or off the rule switches, an optimal rule base can be obtained. An application example will be given.

This paper presents a fuzzy-tuned neural network, which is trained by the genetic algorithm (GA). The fuzzy-tuned neural network consists of a neural-fuzzy network and a modified neural network. In the modified neural network, a novel neuron model with two activation functions is employed. The parameters of the proposed network are tuned by GA with arithmetic crossover and non-uniform mutation. Some application examples are given to illustrate the merits of the proposed network.

This paper presents the tuning of the structure and parameters of a neural network using an improved genetic algorithm (GA). It is also shown that the improved GA performs better than the standard GA based on some benchmark test functions. A neural network with switches introduced to its links is proposed. By doing this, the proposed neural network can learn both the input-output relationships of an application and the network structure using the improved GA. The number of hidden nodes is chosen manually by increasing it from a small number until the learning performance in terms of fitness value is good enough. Application examples on sunspot forecasting and associative memory are given to show the merits of the improved GA and the proposed neural network.

This paper presents the learning of neural network parameters using a fuzzy genetic algorithm (GA). The proposed fuzzy GA is modified from the traditional GA with arithmetic crossover and non-uniform mutation. By introducing modified genetic operations, it will be shown that the performance of the proposed fuzzy GA are better than the traditional GA based on some benchmark test functions. Using the fuzzy GA, the parameters of the neural networks can be tuned. An application example on sunspot forecasting is given to show the merits of the proposed fuzzy GA.

This paper presents a proposed neural fuzzy network tuned by genetic algorithm (GA). By introducing a switch to each rule, the optimal number of rules can be learned. The membership functions of the neural fuzzy network are also tuned by GA. After training, the proposed neural fuzzy network is employed to interpret graffiti number inputs and commands for electronic books (eBooks).

This paper presents the interpretation of graffiti commands for electronic books (eBooks). The interpretation process is achieved by training a proposed neural network (NN) with link switches using an improved genetic algorithm (GA). By introducing the switches to the links, the proposed NN can learn the optimal network structure automatically. The structure and the parameters of the NN are tuned by the improved GA, which is implemented by floating point numbers. The processing time of the improved GA is shorter as reflected by some benchmark test functions. Simulation results on interpreting graffiti commands for eBooks using the proposed NN with link switches and the improved GA, are shown.

This paper tackles the control problem of nonlinear systems subject to parameter uncertainties based on the fuzzy logic approach and genetic algorithm (GA). In order to achieve a stable controller, the TSK fuzzy plant model is employed to describe the dynamics of an uncertain nonlinear plant. A fuzzy controller and the corresponding stability conditions are derived. The parameters of the fuzzy controller and the solution to the stability conditions are determined using GA. In order to obtain the optimal performance, the membership functions of the fuzzy controller are obtained automatically by minimizing a defined fitness function using GA.

In this paper, a short-term home daily load forecasting system realized by a neural fuzzy network (NFN) and an improved genetic algorithm (GA) is proposed. It can forecast the daily load accurately with respect to different day types and weather information. It is also shown that the improved GA performs better than a traditional GA on some benchmark test functions. By introducing switches in the links of the NFN, the optimal network structure can be found by the improved GA. The membership functions and the number of rules of the NFN can be generated automatically. Simulation results for a short-term daily load forecast in an intelligent home are given.

Daily load forecasting is essential to improve the reliability of the AC power line data network and provide optimal load scheduling in an intelligent home system. In this paper, a fuzzy-input-neural network forecaster model is proposed. This model combines a fuzzy system and a neural network. It can forecast the daily load accurately with respect to different day types under various variables. In this model, the fuzzy system performs a preprocessing for the neural network, so that the computational demand of the neural network can be reduced. Simulation results on a daily load forecasting will be given. Comparing the proposed algorithm with that of a conventional neural network, it can be shown that the proposed algorithm produces more accurate forecasting results.