ABSTRACT Reservoir rule curves (RCs) are crucial for guiding operators on the optimal water release based on the available water at the start of each month. In the absence of RCs, simulation and optimization techniques can be effectively employed to develop these curves. This study evaluates the performance of various optimization techniques for deriving optimal reservoir RCs for the Zarrineh Rud reservoir using soft computing (SC) algorithms. The algorithms investigated include the genetic algorithm (GA), particle swarm optimization (PSO), and gravitational search algorithm (GSA). To this end, monthly demand and discharge data from 1987 to 2018 were collected. Historical RCs were first simulated using the sequent peak algorithm (SPA), and optimal RCs were subsequently derived through the GA–SPA, PSO–SPA, and GSA–SPA algorithms to minimize water shortages. The results indicated that the GSA–SPA generally improved the time-based (αt) and volume-based (αv) reliability indices by 3 and 2%, respectively, compared to the historical SPA (SPA-Hist). Additionally, simulations with the GSA–SPA significantly reduced the mean annual shortage and total shortage by approximately 8% compared to SPA-Hist. The PSO–SPA ranked second, with a 7.4 and 6.8% reduction in mean annual shortage and total shortage, respectively.

Cutaneous leishmaniasis (CL) remains a significant global public health disease, with the critical distinction and exact detection between responsive and unresponsive cases dictating treatment strategies and patient outcomes. However, image-based methods for differentiating these groups are unexplored. This study addresses this gap by developing a deep learning (DL) model utilizing transfer learning to automatically identify responses in CL lesions. A dataset of 102 lesion images (51 per class; equally distributed across train, test, and validation sets) is employed. The DenseNet161, VGG16, and ResNet18 networks, pretrained on a massive image dataset, are fine-tuned for our specific task. The models achieved an accuracy of 76.47%, 73.53%, and 55.88% on the test data, respectively, with a sensitivity of 80%, 75%, and 100% and specificity of 73.68%, 72.22%, and 53.12%, individually. Transfer learning successfully addressed the limited sample size challenge, demonstrating the models' potential for real-world application. This work underscores the significance of automated response detection in CL, paving the way for treatment and improved patient outcomes. While acknowledging limitations like the sample size, the need for collaborative efforts is emphasized to expand datasets and further refine the model. This approach stands as a beacon of hope in the contest against CL, illuminating the path toward a future where data-driven diagnostics guide effective treatment and alleviate the suffering of countless patients. Moreover, the study could be a turning point in eliminating this important global public health and widespread disease.

Flyrock represents a significant and fundamental challenge in surface mine blasting, carrying inherent risks to humans and the environment. Consequently, accurate prediction, minimization, and identification of the factors influencing flyrock distance are imperative for effective control and mitigation of its destructive consequences. Machine learning and artificial intelligence methodologies have emerged as viable means to predict and simulate in different scientific fields. This study employs Deep Neural Network in conjunction with three optimization algorithms including the JAYA Algorithm, Multi-Verse Optimization Algorithm, and Gravitational Search Algorithm to predict blasting flyrock distance. The developed model consists of a combination of seven input parameters, encompassing both blasting design parameters and rock geomechanical properties. The output of the Deep Neural Networks model is the flyrock distance. For the training and testing of the model, a dataset comprising of 245 blasting records, collected from Songun copper mine, Iran, was utilized. The DNN model yielded an R2 value of 0.96 and an MSE value of 34.11. These results demonstrate the high accuracy and predictive capability of the model. Furthermore, the application of three optimization algorithms resulted in similar optimized parameter values, which minimized flyrock distances.

Fast and relatively accurate determination of the fragment size distribution of a muck-pile is still a challenge in mining operations and the existing measurement methods are inefficient. In this research, a new algorithm to determine fragment size distribution due to blasting was presented, using the image processing technique. In the newly proposed approach, delineating of the fragmented rock particles, as the main core of processing, was carried out, using a convolutional neural network. Two networks were defined and trained by 150 laboratory and 150 field data images. Also, 30 laboratory and 30 field data images were applied to carry out the validation visually, and by using F1-scores. For the two laboratory and field networks and results obtained by Split-Desktop software automatic edge detection on the same images, the F1-scores are equal to (0.98, 0.74) and (0.99, 0.85) respectively. Also, for determination of fragment size distribution by laboratory data network and Split-Desktop software automatic edge detection on the same images, the Root Mean Square Error (RMSE) for F30 and F80 are equal to (0.36, 1.20) and (0.31, 1.24) respectively. These indicate better performance of the proposed approach for both rock edge detection and fragment size distribution over Split-Desktop software automatic edge detection.

Abstract Background subtraction is a crucial task in computer vision that involves segmenting video frames into foreground and background regions. While deep learning techniques have shown promise in this field, existing approaches typically rely on supervised learning and have limited generalization capabilities for unseen video data. Moreover, many of these methods are not suitable for real-time applications due to their offline or partially online nature. This paper introduces ORGRU, an unsupervised, online, and robust deep learning-based framework for background subtraction. ORGRU utilizes a robust version of Gated Recurrent Units (GRUs) to simultaneously estimate and maintain the background model as the low-rank component while calculating the sparse section as the foreground in a fully online manner. The model is iteratively updated in real time with an unsupervised learning algorithm utilizing only the current frame. To evaluate the effectiveness of the proposed approach, we conduct experiments on the LASIESTA dataset, which is a comprehensive, fully-labeled dataset for change detection covering various background subtraction challenges. The experimental results provide both qualitative and quantitative assessments, demonstrating the robustness and superiority of the proposed approach compared to the state-of-the-art methods.

Genetics play a prominent role in the development and progression of malignant neoplasms. Identification of the relevant genes is a high-dimensional data processing problem. Pyramid gravitational search algorithm (PGSA), a hybrid method in which the number of genes is cyclically reduced is proposed to conquer the curse of dimensionality. PGSA consists of two elements, a filter and a wrapper method (inspired by the gravitational search algorithm) which iterates through cycles. The genes selected in each cycle are passed on to the subsequent cycles to further reduce the dimension. PGSA tries to maximize the classification accuracy using the most informative genes while reducing the number of genes. Results are reported on a multi-class microarray gene expression dataset for breast cancer. Several feature selection algorithms have been implemented to have a fair comparison. The PGSA ranked first in terms of accuracy (84.5%) with 73 genes. To check if the selected genes are meaningful in terms of patient's survival and response to therapy, protein-protein interaction network analysis has been applied on the genes. An interesting pattern was emerged when examining the genetic network. HSP90AA1, PTK2 and SRC genes were amongst the top-rated bottleneck genes, and DNA damage, cell adhesion and migration pathways are highly enriched in the network.

Cutaneous leishmaniasis (CL) imposes a major health burden throughout the tropical and subtropical regions of the globe. Unresponsive cases are common phenomena occurred upon exposure to the standard drugs. Therefore, rapid detection, prognosis and classification of the disease are crucial for selecting the proper treatment modality. Using machine learning (ML) techniques, this study aimed to detect unresponsive cases of ACL, caused by Leishmania tropica, which will consequently be used for a more effective treatment modality. This study was conducted as a case-control setting. Patients were selected in a major ACL focus from both unresponsive and responsive cases. Nine unique and relevant features of patients with ACL were selected. To categorize the patients, different classifier models such as k-nearest neighbors (KNN), support vector machines (SVM), multilayer perceptron (MLP), learning vector quantization (LVQ) and multipass LVQ were applied and compared for this supervised learning task. Comparison of the receiver operating characteristic graphs (ROC) and confusion plots for the above models represented that MLP was a fairly accurate prediction model to solve this problem. The overall accuracy in terms of sensitivity, specificity and area under ROC curve (AUC) of MLP classifier were 87.8%, 90.3%, 86% and 0.88%, respectively. Moreover, the duration of the skin lesion was the most influential feature in MLP classifier, while gender was the least. The present investigation demonstrated that MLP model could be utilized for rapid detection, accurate prognosis and effective treatment of unresponsive patients with ACL. The results showed that the major feature affecting the responsiveness to treatments is the duration of the lesion. This novel approach is unique and can be beneficial in developing diagnostic, prophylactic and therapeutic measures against the disease. This attempt could be a preliminary step towards the expansion of ML application in future directions.

Domain adaptation is a method of transfer learning. Domain adaptation has a source domain and target domain with related but different distributions. Unsupervised domain adaptation could be a scenario wherever we've labeled unlabeled target data and source data. In this paper, an incremental adversarial learning method is proposed for unsupervised domain adaptation. In this work, the unknown target labels are predicted and according to these estimated labels, some target data with more similarity to the source data are added to the source data to improve the adaptation between two domains. We use the adversarial discriminative approach as the base unsupervised domain adaptation technique. We do this to handle the large domain shift between the source and target domain distributions. Experimental reports prove that our approach performs much better on several visual domain adaptation tasks.

The goal of license plate recognition (LPR) is to read the license plate characters. Due to image degradation, there are many difficulties in the way of achieving this goal. In this paper, the proposed method recognizes the license plate characters without employing the traditional segmentation and binarization techniques. This method uses a deep learning algorithm and tries to achieve better learning experience by engaging a multi-task learning algorithm based on sharing features. The features of license plate characters are extracted by a deep encoder-decoder network, and transferred to 8 parallel classifiers for recognition. To evaluate the current work, a database of 11,000 license plate images, collected from a currently working surveillance system installed on a dual carriageway, is employed. The proposed method achieved the correct character recognition rate of 96% for 4000 test images that is acceptable in comparison to the competing methods.

The gravitational search algorithm (GSA) is a meta-heuristic optimization algorithm which is inspired by the gravity force. This algorithm uses Newton’s gravity and motion laws to calculate the masses interactions and shows high performance in solving optimization problems. The premature convergence is the common drawback of heuristic search algorithms in high-dimensional problems, and GSA is not an exception. In this paper, a new version of GSA is proposed to improve the power of GSA in exploration and exploitation. The proposed algorithm has both attractive and repulsive forces. In this algorithm, the heavy particles attract some particles and repulse some others, in which the forces are inversely proportional to their distances. For better evaluation, the GSA with both attractive and repulsive forces (AR-GSA) is tested using CEC 2013 benchmark functions and the results are compared with some well-known meta-heuristic algorithms. The simulation results show that AR-GSA can improve the convergence rate, the exploration, and the exploitation capabilities of GSA.

This paper proposes a totally multiplier-less approximate reconstruction cosine modulated filter bank. The prototype filter is designed to obtain a computationally efficient, sharp transition width, finite impulse response (FIR) filter using the frequency response masking (FRM) approach combined with the canonic signed digit (CSD) representation of the coefficients. Meta-heuristic algorithms are deployed to optimize the performance of the filter bank. The algorithms are modified to suit the optimization problem, in which the objective function is non linear and the search space consists of integers

This paper introduces a multimodal emotion recognition system based on two different modalities, i.e., affective speech and facial expression. For affective speech, the common low-level descriptors including prosodic and spectral audio features (i.e., energy, zero crossing rate, MFCC, LPC, PLP and temporal derivatives) are extracted, whereas a novel visual feature extraction method is proposed in the case of facial expression. This method exploits the displacement of specific landmarks across consecutive frames of an utterance for feature extraction. To this end, the time series of temporal variations for each landmark is analyzed individually for extracting primary visual features, and then, the extracted features of all landmarks are concatenated for constructing the final feature vector. The analysis of displacement signal of landmarks is performed by the discrete wavelet transform which is a widely used mathematical transform in signal processing applications. In order to reduce the complexity of derived models and improve the efficiency, a variety of dimensionality-reduction schemes are applied. Furthermore, to exploit the advantages of multimodal emotion recognition systems, the feature-level fusion of the audio and the proposed visual features is examined. Results of experiments conducted on three SAVEE, RML and eNTERFACE05 databases show the efficiency of proposed visual feature extraction method in terms of performance criteria.

An effective method for optimal design of water distribution network (WDN) can significantly benefit to develop commercial software for component sizing. This research investigates Gravitational Search Algorithm (GSA) for pipe cost optimization model problems. GSA is a meta-heuristic (MH) algorithm which makes ease of its applicability to the design of WDNs due to its minimum number of algorithm parameters and requiring least effort in fine-tuning the parameters. Three well-known benchmark networks (Hanoi network, Two-Reservoir network and New York tunnels network) and a real-world WDN located Khorramshahr city in Iran were used. The GSA results were compared with the solutions obtained through various Evolutionary Algorithms. Experimental results show success of GSA in arriving minimum cost solution. GSA achieved to the best so far solution reported for one case (Hanoi network), and it could find the least cost for three other networks compared to the best results of other optimization algorithms. In addition, for two case studies (Hanoi network and Khorramshahr city network) number of function evaluations were less than other algorithms. Further, the study reveals that GSA achieved the maximum number of times the best so far solution and confirming rapid convergence without struck up at local optimum.

Traffic car images suffer immensely from various degrading factors that make it hard to localize license plates. Each license plate localization (LPL) method has its own advantages and disadvantages to extract plates in the images under different circumstances. To have the benefits of different methods, our proposed solution is to employ a combination of four methods including a method based on cascade classifiers and local binary pattern (LBP) features, an edge-based method, a color-based method, and a contrast-based method. Considering the computational complexity, the methods are ordered on the basis of their chances for success. The order of the methods and the parameters are set experimentally in different conditions: day, night, and twilight. Furthermore, to find the plates rapidly, an algorithm is proposed to refine regions of interest (ROIs) and remove unwanted regions. The algorithm is applied in a real automated transport system for plate identification/recognition and tested with 4000 vehicle images taken from a three-lane dual carriageway with a central barrier in the different illumination situations with six cameras. The results are promising in a large database of moving car images. The car license plates have been correctly extracted in 3938 input images (98.45%). The results show that the proposed system is robust for moving cars in outdoor and under different illumination conditions.

Vehicle License Plate Recognition (VLPR) is one of the most important aspects of applying computer techniques in Intelligent Transport Systems (ITS). They face difficulties like shadows effects, non-uniform illumination intensity, and dirty plates. To tackle these problems, this paper proposes a new VLPR system by producing a contrast enhancement method, a background removal method, and a binarization method. After binarization, an OCR method using artificial neural network (ANN) reads the plate characters. The performance of the proposed system is tested on 4 k Iranian vehicle license plate images. The proposed method causes the correct recognition rate of 91.2%. The results obtained in comparison to those of well-known methods show that the proposed system is robust for moving cars in outside environment and under different illumination conditions.

Gravitational Search Algorithm (GSA) is an optimization method inspired by the theory of Newtonian gravity in physics. Till now, many variants of GSA have been introduced, most of them are motivated by gravity-related theories such as relativity and astronomy. On the one hand, to solve different kinds of optimization problems, modified versions of GSA have been presented such as continuous (real), binary, discrete, multimodal, constraint, single-objective, and multi-objective GSA. On the other hand, to tackle the difficulties in real-world problems, the efficiency of GSA has been improved using specialized operators, hybridization, local search, and designing the self-adaptive algorithms. Researchers have utilized GSA to solve various engineering optimization problems in diverse fields of applications ranging from electrical engineering to bioinformatics. Here, we discussed a comprehensive investigation of GSA and a brief review of GSA developments in solving different engineering problems to build up a global picture and to open the mind to explore possible applications. We also made a number of suggestions that can be undertaken to help move the area forward.

Technological progresses in the gas sensor fields provide the possibility of designing and construction of Electronic nose (E-nose) based on the Biological nose. E-nose uses specific hardware and software units; Sensor array is one of the critical units in the E-nose and its types of sensors are determined based on the application. So far, many achievements have been reported for using the E-nose in different fields of application. In this work, an E-nose for handling multi-purpose applications is proposed, and the employed hardware and pattern recognition techniques are depicted. To achieve higher recognition rate and lower power consumption, the improved binary gravitational search algorithm (IBGSA) and the K-nearest neighbor (KNN) classifier are used for automatic selecting the best combination of the sensors. The designed E-nose is tested by classifying the odors in different case studies, including moldy bread recognition in food and beverage field, herbs recognition in the medical field, and petroleum products recognition in the industrial field. Experimental results confirm the efficiency of the proposed method for E-nose realization.

Today, green cloud computing has been concerned due to the reduction of environmental impacts. One of the criteria that has been emphasized in green cloud computing is energy consumption of data centers. One way to reduce energy consumption, which we is emphasized in this paper, is tasks scheduling management and consolidation of virtual machines. In this paper, an algorithm is presented to manage both tasks scheduling and load balancing. This algorithm, called the Shuffled Frog-Leaping provides a significant improvement against other existing models in terms of energy consumption and migration of virtual machines using memory, collaboration and sharing information among frogs, high convergence speed and better flexibility against local optimum problem. In this paper, the dynamic resource management is based on the consolidation of virtual machines and is implemented according to service level agreement by the proposed method. The difference between this method and other existing methods is that it shows improvement of time, speed and accuracy of convergence parameters. Experimental results show that the proposed method outperforms existing ones in terms of energy consumption, number of virtual machine migrations and service level agreement violation.

Harmony search (HS) is a meta-heuristic search algorithm which tries to mimic the improvisation process of musicians in finding a pleasing harmony. In recent years, due to some advantages, HS has received a significant attention. HS is easy to implement, converges quickly to the optimal solution and finds a good enough solution in a reasonable amount of computational time. The merits of HS algorithm have led to its application to optimization problems of different engineering areas. In this chapter, the concepts and performance of HS algorithm are shown and some engineering applications are reviewed. It is observed that HS has shown promising performance in solving difficult optimization problems and different versions of this algorithm have been developed. In the next years, it is expected that HS is applied to more real optimization problems.