جستجوی مقالات مرتبط با کلیدواژه « Neural network » در نشریات گروه « فناوری اطلاعات »

Nowdays, organizations generate a significant volume of log files that require processing for condition checking, debugging, and anomaly resolution. Outsourcing such processing is not suitable due to the need for real-time processing and security maintenance. Given the multitude of different software and services, organizations face a substantial volume of production logs that should be processed rather than deleted or ignored. In the traditional approach, experts manually check the logs daily. This, on one hand, slows down the process, increases the time and inaccuracy, and, on the other hand, results in a high hiring cost due to the need for an expert force. This article introduces a solution that employs generative neural networks to establish a local structure for log analysis within the organization. The process involves retrieving and parsing text files from various sectors, segmenting them into manageable portions, embedding them, and storing them in a vector database. In this structure, a trained individual without special expertise can quickly access necessary information using appropriate prompts from a local language model available at any time. As a result, three overarching goals are achieved: maintaining security, increasing the speed of analysis, and reducing human resource costs.

Contemporary machine learning models, like deep neural networks, require substantial labeled datasets for proper training. However, in areas such as natural language processing, a shortage of labeled data can lead to overfitting. To address this challenge, data augmentation, which involves transforming data points to maintain class labels and provide additional valuable information, has become an effective strategy. In this paper, a deep reinforcement learning-based text augmentation method for sentiment analysis was introduced, combining reinforcement learning with deep learning. The technique uses Deep Q-Network (DQN) as the reinforcement learning method to search for an efficient augmentation strategy, employing four text augmentation transformations: random deletion, synonym replacement, random swapping, and random insertion. Additionally, various deep learning networks, including CNN, Bi-LSTM, Transformer, BERT, and XLNet, were evaluated for the training phase. Experimental findings show that the proposed technique can achieve an accuracy of 65.1% with only 20% of the dataset and 69.3% with 40% of the dataset. Furthermore, with just 10% of the dataset, the method yields an F1-score of 62.1%, rising to 69.1% with 40% of the dataset, outperforming previous approaches. Evaluation on the SemEval dataset demonstrates that reinforcement learning can efficiently augment text datasets for improved sentiment analysis results.

Integration and diversity of IOT terminals and their applicable programs make them more vulnerable to many intrusive attacks. Thus, designing an intrusion detection model that ensures the security, integrity, and reliability of IOT is vital. Traditional intrusion detection technology has the disadvantages of low detection rates and weak scalability that cannot adapt to the complicated and changing environment of the Internet of Things. Hence, one of the most widely used traditional methods is the use of neural networks and also the use of evolutionary optimization algorithms to train neural networks can be an efficient and interesting method. Therefore, in this paper, we use the PSO algorithm to train the neural network and detect attacks and abnormalities of the IOT system. Although the PSO algorithm has many benefits, in some cases it may reduce population diversity, resulting in early convergence. Therefore,in order to solve this problem, we use the modified PSO algorithm with a new mutation operator, fuzzy systems and comparative equations. The proposed method was tested with CUP-KDD data set. The simulation results of the proposed model of this article show better performance and 99% detection accuracy in detecting different malicious attacks, such as DOS, R2L, U2R, and PROB.

With the accelerated development of Internet finance, electronic funds transfer, and the rapid growth of credit card activity, credit cards play a very important role in every area of ​​life today. There are some risks in this regard that are considered serious threats to both issuers and cardholders. The increasing number of fraudulent credit card transactions forged credit cards and fraudulent use of expired credit cards have led to increased losses. Therefore, finding fraud detection techniques accurately and quickly has become an important topic in current investigations. In this study, after normalizing and reducing the dimensionality of the data using the PCA algorithm, we used the modified perceptron neural network and the grasshopper algorithm to classify the data. In this study, we use the grasshopper algorithm to adjust the weights and biases of neural networks. In the end, we were able to achieve 99.20% accuracy.

The growing number of information on the web and the addition of different web pages and websites to this space has made users face problems. These problems appear to users when users are trying to obtain information on a particular topic, and finding all the pages that are suggested to them is a difficult and time consuming process. In the current research, a profile is first created based on the behavioral characteristics of users at different sessions that result from web server logs. These include things like the frequency of user page views, the length of time the user has been on different pages, and the date the page was viewed. We then group them using the clustering method, then fuzzy inference system, extract the fuzzy rules according to the interests of the users and their clusters, and after obtaining the users’ movement patterns, they Insertneural network into vector format Other tools such as bio-algorithms can be useful by obtaining optimal parameters in optimizing predictions and increasing accuracy in fuzzy neural network. The evaluation criteria in this study is accuracy.

In this study, an alternative method is proposed based on recursive deep learning with limited steps and prepossessing, in which the data is divided into A unit classes in order to change a long short term memory and solve the existing challenges. The goal is to obtain predictive results that are closer to real world in COVID-19 patients. To achieve this goal, four existing challenges including the heterogeneous data, the imbalanced data distribution in predicted classes, the low allocation rate of data to a class and the existence of many features in a process have been resolved. The proposed method is simulated using the real data of COVID-19 patients hospitalized in treatment centers of Tehran treatment management affiliated to the Social Security Organization of Iran in 2020, which has led to recovery or death. The obtained results are compared against three valid advanced methods, and are showed that the amount of memory resources usage and CPU usage time are slightly increased compared to similar methods  and the accuracy is increased by an average of 12%.

This paper presents a vision-based drone detection method. There are a number of researches on object detection which includes different feature extraction methods – all of those are used distinctly for the experiments. But in the proposed model, a hybrid feature extraction method using SURF and GLCM is used to detect object by Neural Network which has never been experimented before. Both are very popular ways of feature extraction. Speeded-up Robust Feature (SURF) is a blob detection algorithm which extracts the points of interest from an integral image, thus converts the image into a 2D vector. The Gray-Level Co-Occurrence Matrix (GLCM) calculates the number of occurrences of consecutive pixels in same spatial relationship and represents it in a new vector- 8 × 8 matrix of best possible attributes of an image. SURF is a popular method of feature extraction and fast matching of images, whereas, GLCM method extracts the best attributes of the images. In the proposed model, the images were processed first to fit our feature extraction methods, then the SURF method was implemented to extract the features from those images into a 2D vector. Then for our next step GLCM was implemented which extracted the best possible features out of the previous vector, into a 8 × 8 matrix. Thus, image is processed in to a 2D vector and feature extracted from the combination of both SURF and GLCM methods ensures the quality of the training dataset by not just extracting features faster (with SURF) but also extracting the best of the point of interests (with GLCM). The extracted featured related to the pattern are used in the neural network for training and testing. Pattern recognition algorithm has been used as a machine learning tool for the training and testing of the model. In the experimental evaluation, the performance of proposed model is examined by cross entropy for each instance and percentage error. For the tested drone dataset, experimental results demonstrate improved performance over the state-of-art models by exhibiting less cross entropy and percentage error.

Today, large-scale vehicles are scattered in different parts of the city and therefore need to be controlled by programmed systems. Applications of these systems include traffic control, urban planning, driverless vehicles, parking lot management by announcing the arrival of a vehicle, detecting stolen or offending vehicles, and so on. Due to challenges such as the multiplicity of objects in the image, weather conditions, different colors and designs of the type of vehicles and very diverse images from different angles of a vehicle in the section identifying the type of vehicles in the photo, Films, moving images, etc. have led to a variety of research, and in this article we will examine some of the techniques.

The neural network learning vector quantization can be understood as a special case of an artificial neural network, more precisely, a learning-based approach - winner takes all. In this paper, we investigate this algorithm and find that this algorithm is a supervised version of the vector quantization algorithm, which should check which input belongs to the class (to update) and improve it according to the distance and class in question. To give. A common problem with other neural network algorithms is the speed vector learning algorithm, which has twice the speed of synchronous updating, which performs better where we need fast enough. The simulation results show the same problem and it is shown that in MATLAB software the learning vector quantization simulation speed is higher than the self-organized neural network.