In this paper, a novel multiobjective optimization model is presented for selecting optimal locations in the water distribution Network (WDN) with the aim of installing pressure loggers. The pressure data collected at optimal locations will be used later on in the calibration of the proposed WDN model.Objective functions consist of maximization of calibrated model prediction accuracy and minimization of the total cost for sampling design. In order to decrease the model run time, an optimization model has been developed using multiobjective genetic algorithm and adaptive neural Network (MOGAANN).Neural Networks (NNs) are initially trained after a number of initial GA generations and periodically retrained and updated after generation of a specified number of full model-analyzed solutions. Trained NNs are replaced with the fitness evaluation of some chromosomes within the GA progress. Using cache prevents objective function evaluation of repetitive chromosomes within GA.Optimal solutions are obtained through pareto-optimal front with respect to the two objective functions. Results show that jointing NNs in MOGA for approximating portions of chromosomes’ fitness in each generation leads to considerable savings in model run time and can be promising for reducing run-time in optimization models with significant computational effort.

Every woman is at risk of ovarian cancer; about 90 percent of women who develop ovarian cancer are above 40 years of age, with the high number of ovarian cancers occurring at the age of 60 years and above. Early and correct diagnosis of ovarian cancer can allow proper treatment and as a result reduce the mortality rate. In this paper, we proposed a hybrid of Synthetic Minority Over-sampling Technique (SMOTE) and Artificial Neural Network (ANN) to diagnose ovarian cancer from public available ovarian dataset. The dataset were firstly preprocessed using SMOTE before employing Neural Network for classification. This study shows that performance of Neural Networks in the cancer classification is improved by employing SMOTE preprocessing algorithm to reduce the effect of data imbalance in the dataset. To justify the performance of the proposed approach, we compared our results with the standard neural Network algorithms. The performance measurement evaluated was based on the accuracy, F-measure, Recall, ROC Area Margin Curve and Precision. The results showed that SMOTE+MLP (with above 96% accuracy) performed better than SMOTE+RBF and standard RBF and MLP.

Selection of proper number and location of boreholes are major tasks in design of exploration Network. Boreholes normally drilled for insitue testing and sampling. A challenging issue in sampling is determination of number and location of samples. The logic of optimum sampling is sequential. This work presents a comparison between single and multiple stages sampling in Semilan dam site. First, the single-stage sampling was tested; using geotechnical data (Lugeon and RQD) of 23 boreholes and indicator function. Based on indicator functions, Lugeon parameter divided to four and RQD to three indexes. The other application of indicator functions, is converting qualitative to quantitative data in order to define indicators of various structures and lithology of the studied dam site. Then, indicator variograms of each parameter, in various directions, were calculated and their variogram parameters were extracted. Based on the average kriging error, Lugeon, RQD and lithology index were also divided into four indexes. Consequently, the function for locations of additional drillings, based on their relation with risk and uncertainty is defined. Function for location of additional drillings is found to be equal to Lugeon multiply to estimation error multiply to index of dam structures; divided by RQD multiply by lithology index.Next, two stages sampling was studied starting with 12 boreholes, according to sample density.All the process was repeated once again and according to function of additional drilling locations, 8 boreholes were selected for the second level or phase of drilling. Based on kriging and neural Network estimator, and using normalized data, desired parameters were estimated.In the present study, kriging variance and estimation error in two stages sampling decreased more, compare to one stage sampling, although the numbers of boreholes decreased from 23 to 20. Reduction of estimation error is related to the layout of the first stage boreholes between the second stage ones.

With the intricate interplay between clinical and pathological data in coronary heart disease (CHD) diagnosis, there is a growing interest among researchers and healthcare providers in developing more accurate and reliable predictive methods. In this paper, we propose a new method entitled the robust artificial neural Network classifier (RANNC) technique for the prediction of CHD. The dataset CHD in this paper has imbalanced data, and in addition, it has some outlier values. The dataset consists of information related to 4240 samples with 16 attributes. Due to the presence of outliers, a robust method has been used to scale the dataset. On the other hand, due to the imbalance of CHD data, three data balancing methods, including Random Over sampling (ROS), Synthetic Minority Over sampling Technique (SMOTE), and Adaptive Synthetic sampling (ADASYN) approaches, have been applied to the CHD data set. Also, six artificial intelligence algorithms, including LRC, DTC, RFC, KNNC, SVC, and ANN, have been evaluated on the considered dataset with criteria such as precision, accuracy, recall, F1-score, and MCC. The RANNC, leveraging ADASYN to address data imbalance and outliers, significantly improved CHD diagnostic accuracy and the reliability of healthcare predictive models. It outperformed other artificial intelligence methods, achieving precision, accuracy, recall, F1-score, and MCC scores of 95.57%, 96.90%, 99.70%, 97.59%, and 93.42%, respectively.

Background and Objectives: Subsampling methods allow sampling signals at rates much lower than Nyquist rate by using low-cost and low-power analog-to-digital converters (ADC). These methods are important for systems such as sensor Networks that the cost and power consumption of sensors are the core issue in them. The Chinese remainder theorem (CRT) reconstructs a large integer (input frequency) from its multiple remainders (aliased or under-sampled frequencies), which are produced from under-sampling or integer division by several smaller positive integers. sampling frequencies can be reduced by approaches based on CRT.Methods: The largest dynamic range of a generalized Chinese remainder theorem for two integers (input frequencies) has already been introduced in previous works. This is equivalent to determine the largest possible range of the frequencies for a sinusoidal waveform with two frequencies which the frequencies of the signal can be reconstructed uniquely by very low sampling frequencies. In this study, the largest dynamic range of CRT for any number of integers (any number of frequencies in a sinusoidal waveform) is proposed. It is also shown that the previous largest dynamic range for two frequencies in a waveform is a special case of our proposed procedure. Results: A procedure for multiple frequencies detection from reminders (under-sampled frequencies) is proposed and maximum tolerable noises of under-sampled frequencies for unique detection is obtained. The numerical examples show that the proposed approach, in some cases, can gain 11.5 times higher dynamic range than the conventional methods for a multi-sensor under-sampling system.Conclusion: Other studies introduced the largest dynamic range for the unique reconstruction of two frequencies by CRT. In this study, the largest dynamic ranges for any number of frequencies are investigated. Moreover, tolerable noise is also considered.