convolutional neural network (cnn)

One of the biometric detection methods is to identify people based on speech signals. The implementation of a speaker identification (SI) system can be done in many different ways, and recently, many researchers have been focusing on using deep neural networks. One of the types of deep neural networks is recurrent neural networks, where memory and recurrent parts are handled by layers such as LSTM or Gated Recurrent Unit (GRU). In this paper, we propose a new structure as a classifier in the speaker identification system, which significantly improves the recognition rate by combining a convolutional neural network with two layers of GRU (CNN+ GRU). MFCC coefficients that have been extracted as cell arrays from each period of Pt speech will be used as sequence vectors for the input of proposed classifier.  The performance of the SI system has improved in comparison to basic methods according to experiments conducted on two databases, LibriSpeech and VoxCeleb1. When Pt is longer, the system performs better, so that on the LibriSpeech database with 251 speakers, recognition accuracy is equal to 92.94% for Pt=1s, and it rises to 99.92% for Pt=9s. The proposed CNN+GRU classifier has a low sensitivity to specific genders, which can be said to be almost zero.

A brain-computer interface (BCI) is a form of assistive technology that facilitates communication between users and machines by interpreting brain signals. The P300 wave, an event-related potential (ERP) in oddball paradigms, is generated approximately 300 milliseconds after the presentation of the target stimulus selected by the user in the brain. Accurate recognition of these waves in a P300 spelling system enables the user to write letters. Classification of P300 waves in an EEG-based spelling system faces several challenges, including accurate detection of P300 waves and handling the high dimensionality of these signals. This study presents a Convolutional Deep Learning Framework (CDLF) for character recognition using EEG signals. The proposed model uses CNN with a one-dimensional kernel to extract features over time. The proposed model was applied to two public datasets: BCI Competition III dataset II and BCI Competition II dataset IIb. The proposed model showed an average character recognition rate of 95% at epoch 15 for BCI Competition III dataset II and 100% at epoch 15 for BCI Competition II dataset IIb, without using any feature and channel selection methods before classification. The proposed model is promising for brain-computer interface classification applications in the spelling domain.

This study utilizes deep-learning models for stock price prediction, focusing on data from five companies listed on the Tehran Stock Exchange over the period 2001 to 2022. Five models are employed, including two hybrid models and three single models. The hybrid CNN-LSTM model serves as the primary model, with its predictive accuracy compared against the other four models. Results indicate that the CNN-LSTM model demonstrates superior performance relative to the others, although the CNN-GRU hybrid model also yields satisfactory results. Interestingly, among the single models, the CNN model surpasses both the LSTM and GRU models, defying initial expectations. The accuracy of the models is notably impacted by factors such as volatility, which increases uncertainty. This research, which exclusively relies on technical indicators, suggests that achieving optimal results hinges not only on selecting the right neural network but also on determining the appropriate number of layers in each model. Overall, the CNN-LSTM model delivers the best performance across four of the five stocks, with the CNN-GRU model slightly outperforming it for one stock. Among the single models, the CNN model consistently outperforms the others.

Misdiagnosis of skin diseases is a common occurrence. Psoriasis is a skin disease that has many similarities with other diseases, and its incorrect diagnosis causes many problems in the treatment process. Misdiagnosis of this disease causes doctors to face problems during treatment. The lack of images of the disease and the database of skin diseases reduces the diagnosis and the coordination of diagnostic methods, therefore, diagnosis using different images is very useful.
Today, diagnosis methods using deep features in medical images have received much attention.
Artificial intelligence is one of the automatic methods of diagnosis. These methods can detect new data entering the system and keep it in memory. Therefore, in this article, two different groups of data have been identified using deep features based on artificial intelligence.
In this method, the data of the first group in the form of training and testing and the data of the second group are studied gradually. If they are correctly identified, the next 0.1 chunks of data enter the network without testing. If they are wrongly recognized, they enter the training section and this reduces the training process. In this work, by training 20% of the data, i.e. the first 10% and the fourth 10%, there was no need for training because the accuracy was not less than98%.
In this article, deep features of images were first extracted using convolutional neural network, and then psoriasis and eczema were diagnosed with average accuracy of98.3%and sensitivity of 97.9% in skin images using artificial intelligence.

In recent years, emotion recognition as a new method for human-computer interaction has attracted the attention of researchers. Automatic speech emotion recognition has become one of the practical methods to increase engagement in most industries. It is expected that emotion recognition based on audio information can result in better accuracy. The purpose of this article is to present an efficient method for recognizing emotional states from speech signals, based on a new cognitive model. Due to the importance of the topic, this article presents an efficient method for recognizing emotional states from speech signals based on a mixed deep learning and cognitive model called SOAR. To implement each part of this model, two main steps have been introduced. The first step is reading the video and converting it to images and preprocessing it. The next step is to use the combination of convolutional neural network (CNN) and learning automata (LA) to classify and detect the rate of facial emotional recognition. The reason for choosing CNN in our model is that no dimension is removed from the speech signal and considering the temporal information in dynamic speech leads to more efficient and better classification. Also, the training of the CNN network in calculating the backpropagation error is adjusted by LA so that the efficiency of the proposed model is increased and the working memory part of the SOAR model can be implemented. In the proposed model, audio databases available in the field of multimodal emotion recognition eNTERFACE' 05 and SAVEE have been used for various experiments. The recognition accuracy of the presented model in the best case from eNTERFACE' 05 and SAVEE databases is equal to 85.3% and 84.5%, respectively.

Deep convolutional neural networks (CNNs) have attained remarkable success in numerous visual recognition tasks. There are two challenges when adopting CNNs in real-world applications: a) Existing CNNs are computationally expensive and memory intensive, impeding their use in edge computing; b) there is no standard methodology for designing the CNN architecture for the intended problem. Network pruning/compression has emerged as a research direction to address the first challenge, and it has proven to moderate CNN computational load successfully. For the second challenge, various evolutionary algorithms have been proposed thus far. The algorithm proposed in this paper can be viewed as a solution to both challenges. Instead of using constant predefined criteria to evaluate the filters of CNN layers, the proposed algorithm establishes evaluation criteria in online manner during network training based on the combination of each filter’s profit in its layer and the next layer. In addition, the novel method suggested that it inserts new filters into the CNN layers. The proposed algorithm is not simply a pruning strategy but determines the optimal number of filters. Training on multiple CNN architectures allows us to demonstrate the efficacy of our approach empirically. Compared to current pruning algorithms, our algorithm yields a network with a remarkable prune ratio and accuracy. Despite the relatively high computational cost of an epoch in the proposed algorithm in pruning, altogether it achieves the resultant network faster than other algorithms.

Today, electricity theft is one of the main challenges for energy distribution and transmission companies around the world. Early detection of abnormal consumers can prevent security and financial losses. Extensive research studies have been done to detect electricity theft by analyzing customer consumption patterns. Today, one of the most widely used methods is convolutional neural networks (CNNs). These networks contain a large number of hyper-parameters.  The accuracy of these networks is low in most studies due to the lack of attention to the adjustment of these hyper-parameters.  Network accuracy and achieving a robust learning model are influenced by the optimal adjusting of these hyper-parameters, which requires exploring a complex and large search space. Meta-heuristic-based search methods are suitable for solving these problems. Therefore, the main contribution of this paper is to use the high ability of the cheetah optimization algorithm (CHOA) to optimally extract CNN hyper-parameters. In this paper, in order to balance the dataset, abnormal samples are created using artificial attacks and added to the dataset. Also, in order to increase the accuracy of the network, abnormal data are clustered using the CHOA algorithm. ISSDA dataset is used to test and evaluate the results. Based on the results obtained and comparing them with the other works, it was proved that the proposed framework with high accuracy identifies abnormal consumers.

Signature identification plays an important role in many areas such as banking, administrative and judicial systems. For this purpose, in this paper, an automatic intelligent framework is developed by combining a deep pre-trained network with a recurrent neural network. The results of the proposed model were evaluated on several valid datasets and collected datasets. Since there was no suitable Persian signature dataset, we collected a Persian signature dataset based on US ASTM guidelines and standards, which can be very effective and profound for deep approaches. Due to the very promising results of the proposed model in comparison with recent studies and conventional methods, to evaluate the resistance of the proposed model to different noises, we added Gaussian Noise, Salt and Pepper Noise, Speckle Noise, and Local var Noise in different SNRs to the raw data. The results show that the proposed model can still be resistant to a wide range of SNRs; So at 15 dB, the accuracy of the proposed method is still above 90%.

Diagnosing benign and malignant glands in thyroid ultrasound images is considered as a challenging issue. Recently, deep learning techniques have significantly resulted in extracting features from medical images and classifying them. Convolutional networks ignore the hierarchical structure of entities within images and do not pay attention to spatial information as well as the need for a large number of training samples. Capsule networks consist of different hierarchical capsules equivalent to the same layers in the CNN neural network. This study tried to extract textural features using a deep learning model based on a capsule network. Thyroid ultrasound images were given to the capsule network as input data, and finally the features learned in the capsule network were used to teach the Support Vector Machine classifier, in order to diagnose thyroid cancer. Experimental results showed that the proposed method with 98% accuracy has achieved better results compared to convolutional networks.