نشریه سیستم های پردازشی و ارتباطی چند رسانه ای هوشمند
سال دوم شماره 2 (پیاپی 4، تابستان 1400)

Parkinson's disease (PD) is one of the most important diseases of the nervous system that occurs due to the degeneration of dopaminergic neurons in the substantia nigra. Since, there is no specific treatment for this disease, timely and clinical diagnosis can play a key role in slowing progression of the disease and providing control strategies to improve the quality of patients' life. In this study, EEG signals analysis has been used as a low-cost, non-invasive, and reliable approach. The proposed method is based on extracting frequency information of EEG signals with Walsh Hadamard transform and calculating a set of features, such as basic statistics, higher-order statistics, impulsive metrics, and entropy from the extracted coefficients. Then, in order to evaluate the proposed method, support vector machine and k nearest neighbor have been used to discriminate patients from the healthy group. The diagnostic ability of the method has been assessed using EEG data of 28 healthy individuals and 28 patients with PD during a reinforcement-learning task. The results showed that the proposed method can detect PD using entropy, support vector machine, and k nearest neighbor with accuracy of 99.95% and 99.98%, respectively.

Deep neural networks have more accuracy and efficiency in solving various problems, but they are vulnerable to Adversarial Examples. Such malicious samples are built to deceive the trained model and to investigate the vulnerability of various neural network models. There are not many successful methods for making these examples in the text domain. In this paper, a robust method based on the model cost function gradient is presented for generating samples and shown that it is possible to replace a small number of words in the original samples with the words that have the most negative impact on classifier's decision to produce new samples similar to the original ones to deceive word level classifier. Finally, using these samples, the accuracy of two trained models were investigated. The method used in this study, by slightly manipulating input samples, has succeeded in reducing the classification accuracy from 86% to less than 10%.

Despite the fact that laboratories are moving towards becoming smart, following the steps of a test protocol and recording new data are still done using paper notes during an experiment, resulting in spreading microbial (e.g. parasites, fungi, viruses, bacteria) and chemical (e.g. such as acrylamide and radioactive materials) contaminations. This study aimed to design an intelligent voice control tool that enables the researchers to read their different tests' steps and record their findings without any need for paper. The hardware includes an electronic board with four operational keys (i.e. next, previous, recording and play) that can be connected to the Android phone using Wi-Fi or Hot spot. In addition, a digital software was written in C Sharp language for Android and then developed using the Xamarian framework. In conclusion, the designed tool is able to reduce the contamination cycle due to the elimination of the paper-taking method in the laboratory.

Digital banking is one of the most important ways of doing financial transactions,
A model has been presented in this study,
considering the cost reduction and saving of time and Internet role in banking services, factors affecting the formation of attitudes,،
The impact of customers' expectations on the final use of Internet Bank among of the Refah Bank Customers. Therefore, factors such as technological acceptance,
perceived justice of services in terms of commitment, trust, and the willingness to use banking services and the Real use of banking services are reviewed.
The population of the present study is customers of all branches in Tehran.
So, 384 questionnaires were distributed among Customers cluster-random sampling method.
The reliability and Validity were examined. In order to analyze the collected data, structural equation modeling was used.
The results of the present study are the effect of quality, perceived justice of service, acceptance of technology and adaptation on attitude, the impact of quality, attitude and compatibility on satisfaction,
The effect of satisfaction on electronic trust, the effect of satisfaction and trust on the desire to use and the impact of the desire to use on Real used.
It is suggested that digital banking will be analyzed in public banks instead of private banks , because public banks have more financial resources to change their background into upgraded one. besides banking customer feel more reliable to invest their investments in public banks.

Using short length codes has hight degree of importance because of I.O.T ubiquitous applications. Of course Deep Learning Methods has acquired state of the art results in object detection, speech processing and some other fields. Meanwhile, Convolutional Neural Network (CNN) models play a fundamental role in Deep Learning methods success. For boosting the accuracy of syndrome based decoding of Low Density Parity Check (LDPC) codes, convolutional neural network was used and for determining the solution of syndrome equations, error pattern recognition method was used. For this purpose a three layered CNN which any layer contains pooling and convolution sub-layer was used. Then output of CNN was applied to Gated Recurrent Unit(GRU). GRU recurrent network with rectified linear unit (ReLU) activation function and node count as three times as codeword length was used. For hyper parameter default value the Tensorflow 2.0 library’s default value was used except some hyper parameters which was altered manually for accuracy improvement reason. comparing between mixed model and bare recurrent model shows that for LDPC code with length of 64, bit error rate drops when mixed model used. In different signal to noise condition, bit error rate drops 0.5 to 0.8 Db toward Maximum Likelihood decoder. Also it is shown that CNNs have potential to improve recurrent networks accuracy when being used together.

Data quality is very crucial for the success Data analysis.The data loaded to the data warehouse must be correct, accurate and must be of very high quality.High quality data in the data warehouse will result in the better analysis and better decision making.So this data quality issues must be addressed before the data is loaded in to the data warehouse. Data cleaning find errors and remove errors.It also detect and deals with data redundancy and data inconsistency. Data cleaning using ETL to ensure quality data in the data warehouse for effective business intelligence. The purpose of data cleansing is to detect so called dirty data (incorrect, irrelevant or incomplete parts of the data) to either modify or delete it to ensure that a given set of data is accurate and consistent with other sets in the system. This research aims to explain and clarify data cleaning method for correcting dirty data. The sample database was defined as the collection of all diseases in the provinces of Zanjan, Elam and Hamedan. In order to solve the problems in the sample database C # and SQL store Procedure was applied. An important part of the results revealed the error after data cleaning was reduced to 0.008 %.