Search

Purpose: Selection of the best stocks for the portfolio as well as allocating the optimal amount of capital per stock in the portfolio are serious challenges in investing in the stock market. The use of machine learning capacities in the process of optimal capital allocation among portfolio assets has received less attention and usually, the same weight is assigned to portfolio stocks or traditional risk assessment methods are used to divide capital between portfolio stocks. The common disadvantage of these methods is that they all use simple and inflexible mechanisms to estimate the performance of a set. The purpose of this paper is to show for the first time, that machine learning can be used to create a more effective mechanism for estimating performance, which leads to a more efficient allocation of capital to portfolio stocks. Methodology: Our proposed framework, uses two predictive models based on machine learning. In the first step, stocks historical information is used in a return forecasting model, then based on the predicted returns, the appropriate stocks of the portfolio are selected. In the second step, a separate forecasting model predicts portfolio returns by taking into account both the forecasted returns in the first model and the expected risk of the stocks. At the end based on the predicted return of the numerous random portfolios, the appropriate weight for each asset is selected. Findings: Comparing the returns of adjusted portfolios with this model and adjusted portfolios with other portfolio optimization methods shows the superiority of the proposed model. Originality/Value: In this paper, by using machine learning models, the process of selecting the appropriate stock of the portfolio and allocating capital among the candidate stocks is done optimally. [ABSTRACT FROM AUTHOR]

The most important cereal crop in the world is rice (Oryza sativa). Over half of the world's population uses it as a staple food and energy source. Abiotic and biotic factors such as precipitation, soil fertility, temperature, pests, bacteria, and viruses, among others, impact the yield production and quality of rice grain. Farmers spend a lot of time and money managing diseases, and they do so using a bankrupt "eye" method that leads to unsanitary farming practices. The development of agricultural technology is greatly conducive to the automatic detection of pathogenic organisms in the leaves of rice plants. Several deep learning algorithms are discussed, and processors for computer vision problems such as image classification, object segmentation, and image analysis are discussed. The paper showed many methods for detecting, characterizing, estimating, and using diseases in a range of crops. The methods of increasing the number of images in the data set were shown. Two methods were presented, the first is traditional reinforcement methods, and the second is generative adversarial networks. And many of the advantages have been demonstrated in the research paper for the work that has been done in the field of deep learning.

Copyright of Iranian Journal of Information Processing & Management is the property of Iranian Information & Documentation Center (IRANDOC) and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

The automatic modulation classification of signals is of great importance in modern communications, especially on cognitive radio. Several methods have been used in this field, the most important of which is the classification of modulation automatically using Deep Learning, where the methods depend on the convolution neural network, which is one of the Deep Learning networks, achieved high accuracy in classifying the modulation, so the proposed network depends on the type of deep learning CNN consisting of four blocks, each block contains a set of symmetric and asymmetric filters. The network also contains Max Pool. In this paper, the features extracted in phase-squaring and polar have been combined for the input, which helps in extending the input, that is, an increase in the features inside the network. It also contributes to improving the accuracy of classifying the higher-order modulation through the Polar plane. The dataset RadioML 2018.01A was adopted, which is used in the most recent research, where 11 types of modulation normal-class: (FM, GMSK, QPSK, BPSK, 0QPSK, AM-SSB-SC, 4ASK, AM-DSB-SC, 16QAM, 8PSK,00K) were taken. A simulation of which can be found in Matlab 2021. The proposed network achieved 100% classification accuracy when the signal-to-noise ratio is greater or equal to 2 dB for 11 types of modulation. The results of the paper were compared with modern networks Baseline network, Visual Geometry Group network, and Residual Neural network. The comparison showed the superiority of the proposed network over these networks, as the proposed network achieved an accuracy equal to 100% at SNR 2 dB while BL achieved an accuracy equal to 72% at SNR 2 dB, RN, and VGG almost reach 93% at SNR 2 dB.

The use of Deep Learning (DL) for identifying false or misleading information, known as fake news, is a growing area of research. Deep learning, a form of machine learning that utilizes algorithms to learn from large data sets, has shown promise in detecting fake news. The spread of fake news can cause significant harm to society economically, politically, and socially, and it has become increasingly important to find ways to detect and stop its spread. This paper examines current studies that use deep learning methods, such as Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs), as well as the multi-model approach, to detect fake news. It also looks at the use of word embedding models to convert text to vector representations and the datasets used for training models. Furthermore, the paper discusses the use of the attention mechanism in conjunction with deep learning to process sequential data.

Background: Brain-computer interface systems provide the possibility of communicating with the outside world without using physiological mediators for people with physical disabilities through brain signals. A popular type of BCIs is the motor imagery-based systems and one of the most important parts in the design of these systems is the classification of brain signals into different motor imagery classes in order to transform them into control commands. In this paper, a new method of brain signal classifying based on deep learning methods is presented. Methods: This cross-sectional study was conducted at Isfahan University of Medical Sciences, School of Advanced Technologies in Medicine, from February 2020 to June 2022. In the pre-processing block, segmentation of brain signals, selection of suitable channels and filtering by Butterworth filter have been done; then data has transformed to the time-frequency domain by three different kinds of mother wavelets including Cmor, Mexicanhat, and Cgaus. In the classification step, two types of convolutional neural networks (one-dimensional and two-dimensional) were applied whereas each one of them was utilized in two different architectures. Finally, the performance of the networks has been investigated by each one of these three types of input data. Results: Three channels were selected as the best ones for nine subjects. To separate 8- 30 Hz, a 5th degree Butterworth filter was used. After finding the optimal parameters in the proposed networks, wavelet transform with Cgauss mother wavelet has the highest percentage in the both proposed architectures. Two-dimensional convolutional neural network has higher convergence speed, higher accuracy and more complexity of calculations. In terms of accuracy, precision, sensitivity and F1-score, two-dimensional convolutional neural network has performed better than one-dimensional convolutional neural network. The accuracy of 92.53%, which is obtained from the second architecture, as the best result, is reported. Conclusion: The results obtained from the proposed network indicate that suitable, and well-designed deep learning networks can be utilized as an accurate tool for data classification in application of motion perception. [ABSTRACT FROM AUTHOR]

Generative models try to obtain a probability distribution that is similar to that of observed data. Two different solutions have been proposed in this regard in recent years: one is to minimize the divergence (distance) between the two distributions by maximizing the variational lower bound, and the other is to implicitly reduce the distance between the two distributions through adversarial processes. One of the problems in generative adversarial networks (GANs) is the mode collapse. Mode collapse is a phenomenon in which, for various inputs, the generative model generates low variety or similar images. This paper tries to provide a solution to the mode collapse problem proposing a novel method called variational generative adversarial networks (VGANs). This method exploits variational autoencoders to initialize GANs. In other words, in addition to maximizing the variational lower bound, it also implicitly reduces the distance between the two distributions. Experimental results show that this method can deal with the mode collapse problem better than the state-of-the-art. Moreover, in the qualitative analysis, according to a survey of 136 people on the authenticity of the generated images, the proposed method can generate images more similar to real ones. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Education & Science is the property of Republic of Iraq Ministry of Higher Education & Scientific Research (MOHESR) and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Sign Language is the most expressive way for communication between deaf and hearing-impaired people, where information is majorly conveyed through the hand gestures. Despite this, one of the most issues that facing people and public domains that are outside the deaf community is understanding sign language, they are need an interpreter to understand this language. Technologies including deep learning and computer vision aid this issue by providing several technical apps and other platforms. In this paper, we propose a real-time sign language detection model using TensorFlow and OpenCV. The main aim of this study is developing model that recognize sign language, and that can be used offline. Through this model, we can detect a common words and sentences that are expressed by hand gestures, like among American and Arabic sign languages like good (thumbs up), bad (thumbs down), thanks, and live long. This study is belonging to the experimental and applicable research. The results of this study were that the model succeed to detect the hand gestures and recognize it with accuracy 100 %.

This research paper proposes a methodology for classifying aerial photographs and lands using deep learning with transfer learning. The study utilizes the Aerial Image Dataset (AID), which contains a diverse set of aerial images with 30 scene classes. The proposed methodology involves data preprocessing, dataset splitting, training images, model selection, model training, and evaluation using performance measures. Three neural network models (ResNet50, VGG19, and EfficientNetB3) are compared, and the best model is selected based on performance metrics such as precision, recall, F1-score, and the confusion matrix. The results show the effectiveness of the proposed methodology in accurately classifying aerial photographs. This indicates that EfficientNetB3 has a higher ability to classify aerial photographs and lands compared to ResNet50 and VGG19. ResNet50 achieved moderate performance with relatively lower precision, recall, and F1-score compared to EfficientNetB3. VGG19, on the other hand, demonstrated the lowest performance across all metrics, showing low precision, recall, and F1-score values. These results can contribute to various applications such as urban planning, real estate development, and land management.

Human posture estimation is a crucial topic in the computer vision field and has become a hotspot for research in many human behaviors related work. Human pose estimation can be understood as the human key point recognition and connection problem. The paper presents an optimized symmetric spatial transformation network designed to connect with single-person pose estimation network to propose high-quality human target frames from inaccurate human bounding boxes, and introduces parametric pose non-maximal suppression to eliminate redundant pose estimation, and applies an elimination rule to eliminate similar pose to obtain unique human pose estimation results. The exploratory outcomes demonstrate the way that the proposed technique can precisely recognize the human central issues, really work on the exactness of human posture assessment, and can adjust to the intricate scenes with thick individuals and impediment. Finally, the difficulties and possible future trends are described, and the development of the field is presented.

In this paper, we adapt AI techniques, especially YOLOv5 algorithm, for the detection and classification of underground buried objects in B-scan ground penetrating radar (GPR) images. We used gprMax toolbox to prepare the dataset and we considered seven models: voids, tunnels, groundwater, rebar and subsurface concrete. Data augmentation was applied to enlarge the Dataset. The recognition results were promising and the algorithm was tested on real GPR images taken from different references. This study shows that YOLOv5 can detect and classify underground targets, which makes it a promising methodology for practical investment to analyze this kind of specific GPR data even with few training samples.

There is growing interest in automating crime detection and prevention for large populations as a result of the increased usage of social media for victimization and criminal activities. This area is frequently researched due to its potential for enabling criminals to reach a large audience. While several studies have investigated specific crimes on social media, a comprehensive review paper that examines all types of social media crimes, their similarities, and detection methods is still lacking. The identification of similarities among crimes and detection methods can facilitate knowledge and data transfer across domains. The goal of this study is to collect a library of social media crimes and establish their connections using a crime taxonomy. The survey also identifies publicly accessible datasets and offers areas for additional study in this area.

Recently, character recognition and deep learning have caught the attention of many researchers. Optical Character Recognition (OCR) usually takes an image of the character as input and generates the identical character as output. The important role that OCR does is to transform printed materials into digital text files. Convolutional Neural Network (CNN) is an influential model that is generous with bright results in optical character recognition (OCR). The state-of-the-art performance which exists in deep neural networks is usually used to handle frequently recognition and classification problems. Many applications are using it, for instance, robotics, traffic monitoring, articles digitization, etc. CNN is designed to adaptively and automatically learn features by using many kinds of layers (convolution layers, pooling layers, and fully connected layers). In this paper we will go through the advantages and recent usage of CNN in OCR and why it’s important to use it in handwritten and printed text recognition and what subjects we can use this technique for. Researchers are progressively using CNN for the machine-printed characters and recognition of handwritten, that is because CNN architectures are suitable for recognition tasks by inputting some images

Deep learning has proved its potential and the vital role it can provide in benefitting and assisting practitioners that use ultrasonography as a tool for diagnosis. However, medical datasets are often very difficult to obtain. Moreover, erroneous or fake training datasets may lead to inaccurate results. This paper is discussing the robustness of deep learning in diagnosing polycystic ovary syndrome (PCOS). Two public online data sets were used to train and test the ability of transfer learning using DenseNet201 architecture in detecting the infected ovaries. The two datasets showed different results although the same model, architecture and parameters were used. Further investigation showed that one of the data sets is extremely erroneous and of misleading information. In conclusion, Convolutional Neural Networks (CNN) and Transfer Learning can be considered as a strong tool to evaluate datasets.

Gastrointestinal (GI) system diseases have increased significantly, where colon and rectum cancer is considered the second cause of death in 2020. Wireless Capsule Endoscopy (WCE) is a revolutionary procedure for detecting Colorectal lesions. It was automatically used to detect the polyps, multiple SB lesions, bleeding, and Ulcer. The acquired video by the WCE can be processed using a Computer-Aided Diagnosis (CAD) system. However, such videos suffer several problems, including burling, high illumination. and distortion. These effects obligate the development of image processing techniques of high accuracy in detection using deep learning-based segmentation. In this paper, a transfer learning-based U-Net was proposed to transfer the knowledge between the medical images in the training phase and the subsequent segmentation using transfer learning to achieve better results and high accuracy results compared to other related studies. The improvement is done by using an algorism written in python code The results showed average segmentation accuracy of 98.67%