In this paper, the challenges of designing and implementing advanced processing circuits in multi core systems have been investigated. Traditional methods of dealing with crosstalk include coding techniques and the use of physical and transistor repeaters, which face limitations such as increased overhead and high complexity. To solve this problem, in this paper, a new approach based on artificial intelligence and deep learning is presented to improve efficiency and reduce crosstalk problems. Using convolutional neural network algorithms, the proposed algorithm is capable of removing inappropriate and harmful patterns in the data and improving the quality of the signals by analyzing and learning them. The results of the simulations show that the proposed method can identify and remove harmful patterns with high accuracy and efficiency, and ultimately lead to an increase in processing speed and a decrease in power consumption in chips. This approach not only improves the performance of processing circuits, but can be used as an effective solution in designing the new generation of multi core chips.

The purpose of speech emotion recognition systems is to create an emotional connection between humans and machine, since recognizing human emotions and goals helps improve interactions between humans and machines. Recognizing emotions through speech has been a challenge for researchers over the past decade. But with advances in artificial intelligence, these challenges have faded. In this study, we took steps to improve the efficiency of these systems by using deep learning methods. In the first step, three-dimensional Convolutional neural networks are used to learn the spectral-temporal Features of speech. In the second step, to strengthen the proposed model, We use the New pyramidal Concatenated three-dimensional Convolutional neural networks, Which is a multi-scale architecture of three-dimensional Convolutional neural networks on input dimensions. Finally, to obtain the ability of learning the spectral-temporal features extracted from the New Pyramidal Concatenated 3D CNN Approach, we used the temporal capsule network, so could be called consider the spatial and temporal relationship of the data. Finally, we named the proposed structure, which is a powerful structure for spectral-temporal feaures, the MSID 3DCNN + Temporal Capsule.The final model has been applied on a combination of two speech and song databases from the RAVDESS database. comparing the results of the proposed model with the conventional models, shows the better performance of our approach. The proposed SER model has achieved an accuracy of 81.77% for six emotional classes by gender.

Since the behavior of people in the videos are in 3D signals format and they are long, it is difficult to search for a specific action. Therefore, a suitable technique in live security videos is required to detect ongoing armed thieves to reduce the occurrence of crime and theft. The innovation of this paper is to provide a rapid and efficient method for detecting guns in frames of images taken from videos without deleting the main points. The hierarchy of object recognition is that in order to extract frames from images derived from videos, the separation algorithm will be applied at a specified frame rate and all images will be placed in a folder. Then, video samples are divided into three categories of training, validation and testing, and using Haar Cascade (HC) classification, the frames of whole body images are extracted and the rest of the backgrounds are removed from the images. The reason for choosing this method is that the HC classification is resistant to rotation of images and also this algorithm has shown good performance compared to complex calculations. Therefore, in our proposed model, we will use this algorithm as a whole body diagnosis. This is done by detecting the Region of Interest (ROI) area by cutting the selected areas, followed by subtracting the background to eliminate unwanted backgrounds. All key points of selection and extraction are stored inside a folder. Finally, all images are sent to 3D convolutional Neural Networks (3DCNNs) to detect weapons in the images. Finally, in order to evaluate the performance of the system in terms of accuracy, it is used with correct positive rate parameters, false positive rate, positive prediction value and false detection rate. As can be seen in the results of the tests, the highest gun detection rate is related to the 3DCNNs model with a detection rate of 96. 1%, followed by the best detection model rate related to YOLO V3 and with a detection rate of 95. 6%.

Condition monitoring and fault diagnosis of large industrial equipment has become very important role nowadays. Powerful artificial intelligent methods can be appropriately used on big data without any further statistical assumption. In this research, two compromising methods including deep neural network and convolutional neural network have been used to classify faults of a laboratory gearbox. Both networks have been used to classify nine faulty classes and one healthy class of the gearbox using vibration signal. The data have been collected at six different load and speed combinations. The measured time domain vibration signal was used as neural network input. The classification accuracy of both methods have been obtained. The effect of challenging parameters such as window size, learning rate and number of extracted features on the classification accuracy have been studied. Finally after the comparison of the results, it was concluded that the accuracy of the convolutional neural network was superior.

Although, speech recognition systems are widely used and their accuracies are continuously increased, there is a considerable performance gap between their accuracies and human recognition ability. This is partially due to high speaker variations in speech signal. Deep neural networks are among the best tools for acoustic modeling. Recently, using hybrid deep neural network and hidden Markov model (HMM) leads to considerable performance achievement in speech recognition problem because deep networks model complex correlations between features. The main aim of this paper is to achieve a better acoustic modeling by changing the structure of deep Convolutional Neural Network (CNN) in order to adapt speaking variations. In this way, existing models and corresponding inference task have been improved and extended. Here, we propose adaptive windows convolutional neural network (AWCNN) to analyze joint temporal-spectral features variation. AWCNN changes the structure of CNN and estimates the probabilities of HMM states. We propose adaptive windows convolutional neural network in order to make the model more robust against the speech signal variations for both single speaker and among various speakers. This model can better model speech signals. The AWCNN method applies to the speech spectrogram and models time-frequency varieties. This network handles speaker feature variations, speech signal varieties, and variations in phone duration. The obtained results and analysis on FARSDAT and TIMIT datasets show that, for phone recognition task, the proposed structure achieves 1. 2%, 1. 1% absolute error reduction with respect to CNN models respectively, which is a considerable improvement in this problem. Based on the results obtained by the conducted experiments, we conclude that the use of speaker information is very beneficial for recognition accuracy.

Automatic recognition and tracking systems of aerial targets are of particular importance in the battle field. These types of systems use visual sensors, have the ability to be installed on various military systems, and are suitable for discovering and tracking low-altitude targets. In this manuscript, a convolutional neural network was designed to recognize the type of aerial targets (cargo, aerobatics, fighter and missile) and then target tracking using a pre-trained network (GoogLeNet) and transfer learning in the form of a region with convolutional neural network was done. The recognition accuracy of aerial targets in the test data set is 96.3%. On the other hand, the overlap value between the actual and predicted bounding box of target in the test data set for cargo and aerobatics plane, fighter and missile is 0.61, 0.66, 0.64 and 0.51, respectively, which shows the desirable accuracy of the developed model for targets tracking in consecutive frames.

With the social media engagement on the rise, the resulting data can be used as a rich resource for analyzing and understanding different phenomena around us. A sentiment analysis system employs these data to find the attitude of social media users towards certain entities in a given document. In this paper we propose a sentiment analysis method for Persian text using Convolutional Neural Network (CNN), a feedforward Artificial Neural Network, that categorize sentences into two and five classes (considering their intensity) by applying a layer of convolution over input data through different filters. We evaluated the method on three different datasets of Persian social media texts using Area under Curve metric. The final results show the advantage of using CNN over earlier attempts at developing traditional machine learning methods for Persian texts sentiment classification especially for short texts.

Introduction: COVID-19 has had a devastating impact on public health around the world. Since early diagnosis and timely treatment have an impact on reducing mortality due to infection with COVID-19 and existing diagnostic methods such as RT-PCR test are prone to error, the alternative solution is to use artificial intelligence and image processing techniques. The overall goal is to introduce an intelligent model based on deep learning and convolutional neural network to identify cases of COVID-19 and pneumonia for the purpose of subsequent treatment measures with the help of lung medical images. Method: The proposed model includes two datasets of radiography and CT-scan. These datasets are pre-processed and the data enhancement process is applied to the images. In the next step, three architectures EfficientNetB4, InceptionV3, and InceptionResNetV2 are used using transfer learning method. Results: The best result obtained for CT-scan images belongs to the InceptionResNetV2 architecture with an accuracy of 99. 366% and for radiology images related to the InceptionV3 architecture with an accuracy of 96. 943%. In addition, the results indicate that CT-scan images have more features than radiographic images, and disease diagnosis is performed more accurately on this type of data. Conclusion: The proposed model based on a convolutional neural network has higher accuracy than other similar models. Also, this method by generating instant results can help in the initial evaluation of patients in medical centers, especially during the peak of epidemics, when medical centers face various challenges, such as lacking specialists and medical staffs.

Summary:Potential field data (gravity and magnetic data) are usually analyzed by employing linear transformations, the spectral method, inversion techniques and analytic signal methods. Nowadays, there are different methods of modeling the gravity data; but each has limitations. One of the limitations of these methods is the assumption of a simple shape for buried structures whereas the actual shape could be entirely different. This study uses cubic units (3D model) to solve this limitation because affords the ability to make any shape for unknown underground structures by arranging these cubics.In this paper, a new method called Forced Neural Networks (FNN) to find the density variation of buried deposits or underground structures in different depth sections by assuming the cubic model is described. The aim of the geological modeling is to determine the shape and location of underground structures in 3-D sections. Here, one neuron network and back propagation algorithm are applied to discover the density difference. In this method, weights of the neurons are assigned as density for each cubic and the activation function has a linear property such that the outputs are the same as the inputs. After using the back propagation, densities for each cubic are updated and the output of the neurons gives the gravity anomaly. Hence, the density differences are found. However, the results of this system are insufficient because non-uniqueness and horizontal locations are constrained; therefore, the value of density difference is set to zero if its value is very close to zero according to the density difference which is obtained from geological features of the region. Otherwise these values are set to the density difference of the geological region after back propagation.Using a forced neural network, after sufficient epoch is applied, fixed values are assigned to the output of the neuron according to the density difference, and this process is continued until the mean square error of the output becomes sufficiently small. The method is used for both noise-free and noise-corrupted synthetic data and, after obtaining satisfactory results for three synthetic data models, this method was used for modeling of the real data.The Dehloran Bitumen map in Iran was chosen as a real data application. The area under consideration is located in the Zagros tectonic zone, west of Iran where we are looking for Bitumen. Layers of Medium-bedded limestone with intermediate marllimestone are the dominant formations in the area and the hydrocarbon zone is one of the most important characteristics of the area. A program was written using the Anomaly modeling method. The final result of this method shows that the deposit begins from the low depth to approximately less than 40 meters. This modeling yeilded satisfactory results for the drilling in the region. The results of the drillings show that the lowest depth of the deposit varies from 7 to 10 meters. This method can easily be applied for gravity, microgravity and magnetic data especially for porphyry deposits.