Background and Objectives: With the increase of population in the world along with the decrease of natural resources, agricultural land and the increase of unpredictable environmental conditions, it causes concerns in the field of food supply, which is one of the serious concerns for all countries of the world. Therefore, the agricultural industry has moved towards smart agriculture. Smart agriculture using the Internet of Things, which uses different types of sensors to collect data (such as temperature, humidity, light, etc.), a communication network to send and receive data, and information systems to manage and analyze data. Smart agriculture deals with a huge amount of data collected from farms, which has fundamental challenges for analysis using old systems such as lack of storage space, processing delay. Computational paradigm is a key solution to solve the problems of time delay, security, storage space management, real-time analysis. Computing paradigms include cloud, fog and edge computing, which by combining each of them in smart agriculture has caused a great transformation in this industry. The purpose of this article is to provide a comprehensive review of the architecture of computing paradigms in smart agriculture applications.Methods: To achieve the goals of this article, the methodology is divided into two parts: article selection and review of the selected articles. The computational paradigms used in the selected articles are from 2019 to 2022. Each selected paper is then reviewed in detail in terms of categories of computing paradigms, architectures, key points, advantages, and challenges.Results: Computational paradigms have significant advantages. Combining these paradigms with each other in a complementary way covers many challenges. The architecture based on the combination of edge-fog-cloud computing is one of the best architectures combined with smart agriculture.Conclusion: By combining computing paradigms and smart agriculture, the challenges based on traditional and old systems are overcome. Combining these paradigms complement each other's challenges.

Introduction In the field of management, the statistics and performance of the deputies and functions of the organization are always of great importance, which requires instant access to the latest status of the system under coverage and minimal forecast of the future situation, to provide quality services Also improve. All of this justifies the existence of an Intelligent statistical system with decision-making capabilities. Methods In this study, we try to create an integrated web-based system in order to electronicize the processes of defining and recording statistical information along with the aggregation of scattered data in the system and finally extract reliable knowledge from this data using modern artificial intelligence methods. In order to make decisions in the fields of health, treatment, medical education as well as health management. The Intelligent management system of North Khorasan University of Medical Sciences has been designed and implemented under the web with the technique of making Webserver software & Learning & data mining in three phases for 48 months in PHP language and MySQL database. Results For this study, the statistical turnover structure of the university was classified into 1058 units. By classification based on the decision tree algorithm and reviewing the records of reports submitted to organizations at the same level and above, we reached 269833 units of data, 376 data types and 3885 data content. In the proposed method, the time cost of data transfer is less than one minute per item, and the required human resources due to the addition of multiple control sensors is 2640 people-hours per year, and significant time and financial savings were made. Conclusions According to the findings, it seems necessary to create a mechanized system for distribution, collection, control, analysis and reporting of statistical indicators that can have high accuracy and time and financial savings.

This research presents an approach based on customer lifetime value (CLV) and artificial neural networks (ANN) to classify bank corporate customers. CLV in banking means how much financial value each customer creates for the bank. The research is of an applied and quasi-experimental type and the statistical population includes 127,672 corporate IDs in the Tejarat Bank Corporate Customer system. Data are collected by scanning customer files over a six-month period. Sampling is by enumeration. K-means algorithms and artificial neural networks are used to cluster customers. Simulations showed that the artificial neural network algorithm provides more accurate results than the K-means algorithm. This approach can effectively classify customers into three clusters. The clusters were reviewed based on the opinions of banking experts; in order to conduct a deeper analysis. Based on the data of Tejarat Bank's corporate customers, these categories were analyzed in terms of CLV. Marketing and sales strategies were developed for each customer cluster. The approach proposed in this research can help banks improve their customer segmentation process and ultimately increase profitability and customer retention.

In traditional speech processing, feature extraction and classification were conducted as separate steps. The advent of deep neural networks has enabled methods that simultaneously model the relationship between acoustic and phonetic characteristics of speech while classifying it directly from the raw waveform. The first convolutional layer in these networks acts as a filter bank. To enhance interpretability and reduce the number of parameters, researchers have explored the use of parametric filters, with the SincNet architecture being a notable advancement. In SincNet's initial convolutional layer, rectangular bandpass filters are learned instead of fully trainable filters. This approach allows for modeling with fewer parameters, thereby improving the network's convergence speed and accuracy. Analyzing the learned filter bank also provides valuable insights into the model's performance. The reduction in parameters, along with increased accuracy and interpretability, has led to the adoption of various parametric filters and deep architectures across diverse speech processing applications. This paper introduces different types of parametric filters and discusses their integration into various deep architectures. Additionally, it examines the specific applications in speech processing where these filters have proven effective.

In this paper, hybrid Intelligent framework presented as a new and efficient method for crude oil price forecasting. This method is developed by applying a systematic integration of GMDH neural networks with GA and Rule-based Expert system (RES) with Web- based Text Mining (WTM) for crude oil price forecasting. In WTM, information about agents that affect on crude oil price outperforms from different sites and degree of these effective agents restores as rules by RES. These rules with long run and short run crude oil price moving averages are modeled by GMDH neural network and results reveal that hybrid Intelligent system improve forecasting results.

Background: Early detection of the breast cancer can significantly increase survival rate among women. Nowadays, researchers aim to automatize fine needle aspiration (FNA), as a simple, non-expensive and non-invasive test for breast cancer diagnosis.Materials & Methods: Intelligent diagnosis of breast cancer consists of 5 steps: fluid extraction from the breast lump, capturing digital microscopic images from the samples, extracting morphological real-valued features from the images, feature selection and designing a pattern recognition system to distinguish between benign and malignant tumors. Using WDBC database (including 569 FNA samples), a novel BPSO-based feature selection method and SVM classifiers an Intelligent breast cancer diagnosis system is developed.Results: Merit of the proposed system is successfully certified on WDBC dataset leading to recognition rate of %100 using only 28 features (in 5 SVM models). The system clearly outperforms previous works in both respects of accuracy and the number of required features.Conclusion: Developing a novel efficient feature selection algorithm can improve both accuracy and speed of Intelligent breast cancer diagnosis systems. In addition to general diagnosis, using feature selection would help physicians discovering abnormalities caused by diseases.

With the electronic taxpayers system being operationalized and the digital storage of tax data developed in Iran, it is now possible to design different models to analyze the available data. There are two main areas that have not been the focus of the fairly limited current studies in this field; one being the parallel optimization of parametric AI models and the other area is the selection of input variable combination. For this reason, in present study, we have used the harmony search (HS) optimization algorithm to do parallel optimization of multilayer perceptron (MLP) neural network parameters and also to find a suitable combination of input variables. In addition to that, the results have been compared with logistic regression results as the core of the system. In the present research, 21 initial input variables are selected for the system based on the survey done on similar studies in the last thirty years and it takes into account the specifications of the tax system in Iran and the opinions of the experts in the field are asked. After running the system on the data from the food and textile sectors and comparing the results from the neural network and logistic regression, we have concluded that neural network can produce more accurate results and the difference is statistically meaningful.

This article will introduce a robust vision system which was implemented on a mobile manipulator. This robot has to find objects and deliver them to pre specified locations. In the first stage, a method which is named color adjacency method was employed. However, this method needs a large amount of memory and the process is very slow on computers with small memories. Therefore since the previous methods, had used statistical method for object detection, the samples for the connectionist were extracted by this method. The obtained neural network is very robust against light changes and can detect objects very quickly. This neural network has the advantage of being extremely simple to implement, and astonishingly quick in practice.