1,720,996 research outputs found
Motor imagery EEG classification using algorithms and machine learning for ALS disease
Get PDFThe cerebrum PC interface brings issues to light of basic ailments influencing the human sensory
system. The human sensory system is a mind boggling organization of neurons with a simple
flagging component. The cerebrum PC interface permits the framework to keep human mind
action to recognize sickness utilizing AI and PC vision. BCI information is alluded to as engine
symbolism. Engine symbolism is non-fixed information that is planned regarding recurrence and
time. In engine picture, recurrence is separated into various sign groups like alpha, beta, gamma,
and delta. Because of the great component of information and clamor relics, band division with
recoded signal is a basic undertaking. The information planning and vector development
processes require information decrease, however the information decrease process loses some
band esteem and can't correct the exact band of sign for the location of ailment and explicit
sickness problems. This review centers around highlight extraction from engine symbolism EEG
information and characterization of EEG information with human sickness
Data gathering in wireless sensor network using multiple mobile sinks and integrated particle swarm optimization with tabu search
Get PDFData gathering is an active research topic for wireless sensor networks WSNs with internet of
things IoTs. For nodes with limited resources, optimal data collection offers effective sensor
data collection at the lowest possible cost and energy consumption. Enabling mobile sinks for
data collection aids in additional cost reduction and improves energy conservation. This study
introduces a unique multi-objective variable length particle swarm optimization for data
gathering in WSNs with various mobile sinks that also incorporates Tabu search. The integrated
algorithm is known as variable length particle swarm optimization with Tabu search (SCPSOTS). It offers the greatest variety of mobile sinks and their related trajectory-defined rendezvous
points (RPs). The evaluation revealed that SCPSO-TS outperformed the benchmarks in terms
of the evaluation criteria
Using deep learning algorithms character-based word generation
No full textBu çalışmanın amacı derin öğrenme algoritmaları kullanılarak karakter tabanlı bir dil modeli geliştirmek ve Türkçe dil bilgisi kurallarına uygun anlamlı kelime üretmektir. Metin üretimi, bilgisayarların insanların iletişim kurma şeklini taklit ederek doğal dilde metin oluşturmak için kullanılmaktadır. Doğal dilde metin verileri yan yana olarak ve zaman sırasına göre sıralanmış birbiriyle ilişkili kelime veya kelime gruplarının oluşturduğu bütünsel bir düzen şeklinde bulunmaktadır. Metin içerisinde bulunan karakterlerin ve kelimelerin sırasının ve anlamının önemi çok büyüktür. Bu nedenle oluşturulacak kelime, cümle ve metin parçalarında karakterler veya kelimeler arasındaki ilişkilerin iyi öğrenilebilmesi için geçmiş bilgileri hatırlayarak öğrenen derin öğrenme algoritmalarının kullanılması gerekmektedir. Derin öğrenme algoritmalarından özyinelemeli yapay sinir ağları geçmiş bilgileri hatırlayarak sıralı örüntüler oluşturmada başarılı sonuçlar verdiği için bu alanda etkili bir şekilde kullanılmaktadır. Çalışmada LSTM, GRU, Kodlayıcı-Kod Çözücü ve dikkat mimarisi kullanılarak Türkçe dil bilgisi kurallarına uygun anlamlı kelimeler üretilmektedir. Çalışmada özellikle Kodlayıcı-Kod Çözücü ve dikkat modeline ağırlık verilmektedir. Nedeni ise diğer modellerin (LSTM/GRU) giriş ve çıkış verilerinin uzunluğu arttıkça karakterler arasındaki ilişkileri hatırlamakta yetersiz kalmalarıdır. Bu sorunların üstesinden gelebilmek için daha uzun girdi ve çıktı dizelerinde daha başarılı sonuçlar veren Kodlayıcı-Kod Çözücü ve Dikkat mimarisi kullanılmaktadır. Çalışmada her bir model 100, 300, 500 epoch değerlerinde sıcaklık örnek alma yönteminin farklı eşik değerlerinde çalıştırılmaktadır. LSTM modeli 100 epoch ve sıcaklık örnek alma yönteminin 0.5 eşik değerinde 91,3% başarı oranı ile en iyi sonucu, GRU modeli 300 epoch ve sıcaklık örnek alma yönteminin 0.5 eşik değerinde 93.8% başarı oranı ile en iyi sonucu, Kodlayıcı-Kod Çözücü ve Dikkat modeli 500 epoch ve sıcaklık örnek alma yönteminin 0.2 eşik değerinde 95.7 % başarı oranı ile en iyi sonucu vermektedir. Çalışma sonunda kodlayıcı-kod çözücü ve dikkat mimarisi kullanılarak oluşturulan dil modelinin anlamlı kelime üretiminde ve daha uzun kelime gruplarının oluşturulmasında diğer modellere (LSTM,GRU) göre daha yüksek başarı elde edildiği görülmektedir. Karakter tabanlı geliştirilen modellerin Türkçe dil bilgisi kurallarına uygun kelime ve kelime grubları üretmede başarılı olduğu görülmektedir.The aim of this study is to develop a character-based language model using deep learning algorithms and to produce meaningful words in accordance with Turkish grammar rules. Text generation is used to create text in natural language by imitating the way computers communicate. In natural language, text data exists side by side and in the form of a holistic arrangement of related words or phrases arranged in chronological order. The order and meaning of the characters and words in the text are of great importance. For this reason, deep learning algorithms that learn by remembering past information should be used in order to learn the relationships between characters or words in words, sentences and text fragments to be created. Recursive artificial neural networks, one of the deep learning algorithms, are used effectively in this field because they give successful results in creating sequential patterns by remembering past information. In the study, meaningful words in accordance with Turkish grammar rules are produced by using LSTM, GRU, Encoder-Decoder and attention architecture. In the study, especially the Encoder-Decoder and attention model are emphasized. The reason is that other models (LSTM/GRU) fail to remember the relationships between characters as the length of the input and output data increases. In order to overcome these problems, the Encoder-Decoder and Attention architecture is used, which gives more successful results in longer input and output strings. In the study, each model is operated at different threshold values of the temperature sampling method at 100, 300, 500 epoch values. LSTM model 100 epoch and temperature sampling method best result at 0.5 threshold with 91.3% success rate, GRU model 300 epoch and temperature sampling method best result with 93.8% success rate at 0.5 threshold, Encoder-Decoder and Attention model gives the best result with a success rate of 95.7% at 500 epochs and 0.2 thresholds of the temperature sampling method. At the end of the study, it is seen that the language model created by using encoder-decoder and attention architecture has higher success in producing meaningful words and creating longer word groups than other models (LSTM, GRU). It is seen that the character-based models are successful in producing words and word groups in accordance with Turkish grammar rules
CNN özellik haritası ve görüntü işleme teknikleri kullanılarak dikiş işleminde kırık dikiş tespit yöntemi
No full textEndüstriyel süreçlerde kalite giderek daha önemli hale geliyor ve maliyetlerin düşürülmesi ve süreç optimizasyonu giderek daha gerekli hale geliyor. Kalite kontrolü, bir süreç için artan üretim ve hatta artan kar sağlar. O halde üretim söz konusu olduğunda en önemli ölçüt olduğu söylenebilir. %100 hatasız bir üretim sürecine sahip olmak son derece zordur. Kusurlar açısından en çok ilgi gören endüstriyel süreçlerden biri de dokuma işlemidir. Bu çalışma, Makine Öğrenimi teknikleri ve ayrıca Dalgacıklar üzerine küresel bir çalışma yapacaktır. Bu çalışma gelecekte yapılacak akademik çalışmalara temel teşkil edebilir. Mevcut çalışmada oluşturulan uygulama aynı zamanda bir bilgisayarlı görme sisteminin, kullanılan sınıflandırıcı algoritmasından, bu durumda Dalgacık Dönüşümünü kullanacak olan özellik çıkarma tekniğine kadar nasıl değişebileceğinin bir örneği olarak hizmet edecektir. Bu çalışmada kumaşlardaki kusurları tespit etme yöntemlerinde en son teknolojiyi araştırıyoruz. Ayrıca veritabanını ve kullanılacak görsel setini de oluşturacağız. Dalgacık Dönüşümü ile görüntüden bilgi çıkarın. Bu soruna en iyi cevabı bulmak için farklı sınıflandırma algoritmalarını test edin. CNN algoritması aracılığıyla sınıflandırıcı algoritmaların performansını artırın. K-katlı çapraz doğrulama tekniğini kullanarak sistemi doğrulayınQuality in industrial processes has become increasingly important and cost reduction and process optimization are becoming increasingly necessary Quality control brings increased production and even increased profits for a process. It can be said, then, that it is the most important metric when it comes to production. It is extremely difficult to have a 100% defect-free manufacturing process. One of the industrial processes that have received much attention regarding defects is the weaving process. The present work will make a global study on Machine Learning techniques and also on Wavelets. This study may serve as a basis for future academic work. The application built in the present work will also serve as an example of how a computer vision system can vary from the classifier algorithm used to the feature extraction technique, which in this case, will use the Wavelet Transform. In this work, we Survey the state of the art in methods of recognizing defects in fabrics. We will also Create the database, as well as the set of images to be used. Extract information from the image with the Wavelet Transform. Test different classification algorithms to find the best answer for this problem. Improve the performance of the classifier algorithms through the CNN algorithm. Validate the system using the k-fold cross-validation techniqu
Early predictive analysis for heart attack identification
Get PDFThis thesis investigates the evolution of an early predictive analysis system for the
identification of heart attacks, with the help of Arduino. The purpose of this research is to
design and develop a reliable and cost-effective mechanism that can accurately identify the
risk of a heart attack before it occurs.
The proposed system utilizes an Arduino microcontroller board to accumulate and process
real-time data from a range of sensors, including heart rate, blood pressure, and other vital
signs. The collected data is then transmitted to a machine learning algorithm, which analyses
the data and speculate the likelihood of a heart attack occurring.
The system is designed to trigger an alarm if the predicted risk of a heart attack exceeds a
predefined threshold. This alarm can be sent to healthcare providers or family members,
allowing for timely intervention and prevention of heart attacks.
The system's reliability and accuracy were evaluated through extensive testing on a dataset
of patient records, which demonstrated a high level of accuracy in identifying individuals at
risk of experiencing a heart attack. The system's low cost and portability make it accessible
to a broader range of users, particularly those in developing countries or with limited
healthcare access. The findings of this research have significant implications for the prevention and treatment
of heart disease, particularly in underserved communities. The system developed in this
research demonstrates the potential of using low-cost and accessible technology such as
Arduino to develop early predictive analysis systems for heart attack identification
Bir metin tasarımı ve simülasyonuoptik kullanarak tanıma algoritmasıkarakter tanıma (OCR) tekniği
No full textElektronik kamera ve belge görüntü güvenliği, ortaya çıkan iyi bilinen gereksinimlerdir. optik karakter tanıma (OCR) ve metin algılama alanında. Bazen çok var Görüntülerdeki bozulmalar ve metnin net olmayan görüntüsü iletilir, bu da kullanılmasını gerektirir. görüntü için işleme teknikleri ve ardından harfler üzerinde optik tanıma. Bu çalışmada sunduğumuz önemsiz dosya görüntüsü dalgalanmalarını telafi etmek için benzersiz bir tek, parametrik olmayan prosedür ideal bir dünyada OCR doğruluğunu artırmayı amaçlamaktadır. Bizim yöntemimiz çok görüntü bozulmasını azaltmak için etkili görüntü geliştirme teknikleri seti. İlk adım, biz aydınlatma farklılıkları ve düzensizliklerle başa çıkmak için parlaklığı ve kontrastı ayarlama yöntemi önerir. görüntü aydınlatmasının dağılımı. İkinci adımda, geliştirilmiş bir gri tonlama dönüştürme kullanıyoruz belge resmimizi gri tonlamaya dönüştürmek için algoritma. Üçüncü adım, Keskinliği Azalt'ı kullanmaktır. Gri tonlamalı görüntüyü keskinleştirmek için maskeleme yöntemi. Son olarak, başlatmak için ideal bir yöntem kullanılır. OCR tanıma için görüntü. Önerilen Sistem genellikle metin algılama düzeyini artırabilir ve OCR doğruluğu. Yaklaşımımızın etkinliğini kanıtlamak için, tam deneyler şurada sunulmaktadır: standart bir veri seti.Electronic camera and document image security are well-known requirements that have appeared in the field of optical character recognition (OCR) and text detection. Sometimes there are many distortions in the images and uncleared image of the text is transmitted, which requires to use of processing techniques for the image and then optical recognition on letters. In this work, we present a unique single, a non-parametric procedure for compensating for junk file image fluctuations aimed at enhancing OCR accuracy in an ideal world. Our method relies on the use of a very effective set of image enhancement techniques to reduce image distortion. The first step, we suggest a method of adjusting brightness and contrast to deal with lighting differences and irregular distribution of image illumination. The second step, we use an improved greyscale conversion algorithm to transform our document image to greyscale. The third step is using the Un-sharp Masking method to sharpen the grayscale image. Finally, an ideal method is used to initialize the image for OCR recognition. The Proposed System can generally enhance the text detection level and OCR accuracy. To prove the efficiency of our approach, full experimentation is presented on a standard dataset
A decision support system based on machine learning for land investment
Get PDFThis research proposes a methodology for classifying aerial photographs and extracting
investible lands using deep learning (DL) 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 image
augmentation, 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, sensitivity, Fmeasure, and the CM. The results observe the effectiveness of the proposed methodology in
accurately classifying aerial photographs and identifying investible. This indicates that
EfficientNetB3 has a higher ability to classify aerial photographs and extract investible lands
compared to ResNet50 and VGG19. ResNet50 achieved moderate performance with
relatively lower precision, sensitivity, and F-measure compared to EfficientNetB3. VGG19,
on the other hand, demonstrated the lowest performance across all metrics, showing low
precision, sensitivity, and F-measure values. These results can contribute to various
applications such as urban planning, real estate development, and land management, where
accurate classification of aerial images is crucial for decision-making processes. Finally, the
future work may involve exploring additional deep
Age and gender detection by face segmentation and modefied CNN algorithm
No full textThe fundamental purpose of the field of research known as "biometrics" is to explore the
development of reliable approaches for identifying individuals using their observable traits.
Examples of biometric identification include both physical and mental traits of an
individual. As a kind of physical identification, fingerprints and facial features may be
compared and analyzed. Human gait has been researched because it has the potential to be
used as a behavioral identifier in computer vision Using a Convolutional Neural Network
(CNN) and a Support Vector Machine, the capacity to estimate a person's gender based on
their stride was explored and investigated (SVM). Throughout our examination of CNN's
potential uses for gait-based gender identification, we will strive for both a high degree of
accuracy and a cheap computational cost. We analyzed and experimented with a variety of
CNN architectures and hyperparameters in this setting
Using cloud computing services in the knowledge sharing process in Iraqi universities
No full textThe study aimed to reveal the effect of cloud computing (CC) in knowledge sharing (KS), and to
achieve the essence of the methodology, a questionnaire was developed to ensure the existence
of a relationship between the study variables. The study sample included (307) questionnaires on
faculty members in Iraqi public universities, and their data was analyzed using the program
(SPSS V.24, Excel 2013).
The results of the study showed that there is a significant effect of (CC) on knowledge sharing
(KS). Therefore, universities need to rely on (CC) to share knowledge. The results of the current
study can contribute to directing the attention of workers in the educational sector towards the
use of (CC) applications to share knowledge in the educational process, Thus, increasing the
chances of universities in achieving their goals and objectives
Prediction of tumor in mammogram images using data mining models
Get PDFInformation Mining (DM) is a cycle that concentrates designs from huge datasets to address
information, with an emphasis on issues like plausibility, value, viability, and versatility.
Information preprocessing includes a few phases, including I information cleaning, (iii)
information combination, (iv) information choice, and (v) information change, before they are
prepared for mining. DM strategies have made huge commitments to a few significant
investigates and fields of conventional sciences, for example, cosmology, high energy physical
science, science, the World Wide Web, Big Data Analytics, superior execution figuring,
distributed computing, medication, and other related spaces. This review looks at the
presentation of grouping calculations by characterizing the result in view of the quantity of
white variety pixels utilizing arrangement calculations. Altogether, ten ascribes are utilized for
characterization. The power of pictures (number of pixels) is viewed as one of the critical
qualities in light of the aftereffects of bunching techniques, and it is incorporated close by the
current nine different characteristics. Truck, SVM, Nave Bayes, and J48 arrangement
strategies are utilized with different exactness measures, for example, FP rate, TP rate, Recall,
Precision, ROC Area, and F-measure
- …
