检索结果-内蒙古大学图书馆

Deep Learning Approach to Diagnose Alzheimer’s Disease through Magnetic Resonance Images

学校读者我要写书评

暂无评论

Deep Learning Approach to Diagnose Alzheimer’s Disease thro...

International Conference on Innovative Computing (ICIC)

作者： Muhammad Faraz Ahmad Shahzad Akbar Syed Al E Hassan Amjad Rehman Noor Ayesha Ripah College of Computing Riphah International University Faisalabad Pakistan Artificial Intelligence & Data Analytics Lab (AIDA) Prince Sultan University Riyadh Saudi Arabia School of Clinical Medicine Zhengzhou University Zhengzhou Henan RP China

ISBN: (纸本)9781665400923

Alzheimer’s disease (AD) is the disease that is neurodegenerative which causes 70 to 80% of cases of dementia worldwide. Even though over the past few years’ research on Alzheimer’s disease has risen, Although, the brain function and structure complexity enables diagnosis still a difficult task. In recent years, Machine Learning approaches have played an essential role in understanding the deep analysis of diseases detection with the help of drug delivery and image processing units in biomedical science. These techniques have enhanced the efficiency and ability to understand complex medical situations. This paper investigates a Convolutional Neural Networks (CNN) based Algorithm of Deep learning and Computer-Aided Diagnosis (CAD) for the Alzheimer’s disease (AD) early diagnosis by differentiating the affected brain and normal brain. Since detection of AD has been a difficult task. However, using CNN-based CAD, we successfully classify the MRI data of AD subjects with an accuracy rate of 97%. This research suggests that Deep Learning algorithms are the most successful methods for classifying clinical MRI subjects by the scale and shift homogeneous distill features extracted by a Convolutional Neural Network. In addition, this approach will help doctors and clinicians to understand and predict more complex systems in the biomedical field.

关键词： Deep learning Magnetic resonance imaging Information filters Feature extraction Classification algorithms Convolutional neural networks Gabor filters

FS-Net: Full Scale Network and Adaptive Threshold for Improving Extraction of Micro-Retinal Vessel Structures

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Getahun, Melaku N. Rogov, Oleg Y. Dylov, Dmitry V. Somov, Andrey Bouridane, Ahmed Hamoudi, Rifat Moscow Russia Center for Data Analytics and Cybersecurity University of Sharjah Sharjah United Arab Emirates Moscow Russia Research Institute for Medical and Health Sciences College of Medicine University of Sharjah Sharjah United Arab Emirates Division of Surgery and Interventional Science Faculty of Medical Science University College London London United Kingdom BIMAI-Lab Biomedically Informed Artificial Intelligence Laboratory University of Sharjah Sharjah United Arab Emirates Bian Que AI Hong Kong

Retinal vascular segmentation, a widely researched topic in biomedical image processing, aims to reduce the workload of ophthalmologists in treating and detecting retinal disorders. Segmenting retinal vessels presents unique challenges;previous techniques often failed to effectively segment branches and microvascular structures. Recent neural network approaches struggle to balance local and global properties and frequently miss tiny end vessels, hindering the achievement of desired results. To address these issues in retinal vessel segmentation, we propose a comprehensive micro-vessel extraction mechanism based on an encoder-decoder neural network architecture. This network includes residual, encoder booster, bottleneck enhancement, squeeze, and excitation building blocks. These components synergistically enhance feature extraction and improve the prediction accuracy of the segmentation map. Our solution has been evaluated using the DRIVE, CHASE-DB1, and STARE datasets, yielding competitive results compared to previous studies. The AUC and accuracy on the DRIVE dataset are 0.9884 and 0.9702, respectively. For the CHASE-DB1 dataset, these scores are 0.9903 and 0.9755, respectively, and for the STARE dataset, they are 0.9916 and 0.9750. Given its accurate and robust performance, the proposed approach is a solid candidate for being implemented in real-life diagnostic centers and aiding ophthalmologists. © 2023, CC BY-NC-ND.

关键词： Medical image processing

Application of Business intelligence Solution Development and Implementation in a Small-Sized Enterprise

学校读者我要写书评

暂无评论

Application of Business Intelligence Solution Development an...

2020 First International Conference of Smart Systems and Emerging Technologies (SMARTTECH)

作者： Hind Fahad Alaskar Tanzila Saba Artificial Intelligence and Data Analytics (AIDA) Lab College of Computer and Information Sciences Prince Sultan University Riyadh Saudi Arabia

ISBN: (数字)9781728174075

ISBN: (纸本)9781728174082

Business intelligence (BI) is an approach to optimize business decisions by utilizing existing information. BI has been supporting the organization to form effective strategic and tactical business strategies. For the past 2 decades, BI hasn't only had a great impact on large enterprises, but also it has aided in forming strong core strategies of small to mid-sized enterprises. This paper details the full BI solution's cycle; planning, design, and development with a focus on small enterprises. Fig. 1 (“System Development Cycle”) summarizes the process followed to develop a BI solution in this paper. The study involves a case study of a fictitious company, created for the sole purpose of applying the suggested BI solution. This case study elaborates on all the phases required to develop a BI solution for each phase. The challenges and benefits of the BI solution are thoroughly considered in this study. To evaluate the functionality of the proposed BI solution, several testing iterations were performed and the results of these iterations are presented at the end of this case study. The paper concludes by highlighting the areas for future research in BI.

关键词： Companies Planning Business intelligence Testing

Covid-19 Detection by using Deep learning-based Custom Convolution Neural Network (CNN)

学校读者我要写书评

暂无评论

Covid-19 Detection by using Deep learning-based Custom Convo...

International Conference on Innovative Computing (ICIC)

作者： Mazhar Javed Awan Muhammad Wasif Imtiaz Muhammad Usama Amjad Rehman Noor Ayesha Hafiz Muhammad Faisal Shehzad University of Management and Technology Lahore Pakistan Artificial Intelligence & Data Analytics Lab (AIDA) CCIS Prince Sultan University Riyadh Saudi Arabia School of Clinical Medicine Zhengzhou University Zhengzhou Henan RP China University of Sargodha Pakistan

ISBN: (纸本)9781665400923

The early diagnosis and treatment of COVID-19 has been a challenge all over the world. It is challenging to manufacture many testing kits and even then, their accuracy rate is very low. Studies carried out recently show that chest x-ray images are of great help in the diagnosis of COVID-19. In this study, we have developed a COVID-19 detection model that by observing the chest x-ray images of the patient, detects that either the patient is affected by COVID-19 or not. The model is developed using a custom Convolutional Neural Network (CNN) that differentiates between COVID-19 and healthy x-ray images so that the patient can be diagnosed and quarantined on time to prevent the spread of the pandemic. We used two different datasets which are publicly available for the training and validation of this model. Upon completion, the proposed model yields an accuracy of almost 98%. Upon further training, our model will be able to be used as a COVID-19 detection tool in hospitals worldwide and will play a vital role in early detection and timely containment of the pandemic.

关键词： COVID-19 Training Pandemics Hospitals Neural networks Convolutional neural networks X-ray imaging

Stock Price Prediction and Portfolio Optimization Using Hidden Markov Model and Quadratic Programming

学校读者我要写书评

暂无评论

Stock Price Prediction and Portfolio Optimization Using Hidd...

Integrated intelligence and Communication Systems (ICIICS), International Conference on

作者： August C. Thio-Ac Daniel D. Cabrales Drexanne E. Calibara Maries V. Madrazo Agustine Christian C. Matawaran Genesis E. Micaller Edmon O. Fernandez Timothy M. Amado Romeo L. Jorda Lean Karlo S. Tolentino Lejan Alfred C. Enriquez Department of Electronics Engineering Technological University of the Philippines Manila Philippines Integrated Research and Training Center Technological University of the Philippines Manila Philippines Intelligent Systems Applications and Circuits Laboratory (ISAAC) Integrated Research and Training Center Technological University of the Philippines Manila Philippines Data Analytics Visualization Information Driven (DAVID) Research Lab Integrated Research and Training Center Technological University of the Philippines Manila Philippines Center for Artificial Intelligence and Nanoelectronics (CAIN) Integrated Research and Training Center Technological University of the Philippines Manila Philippines

This study introduces a decision support system for optimizing stock portfolios. It employs a Hidden Markov Model for 95% accurate stock price forecasts over a five-day period. The system utilizes Quadratic Programming to refine allocation, focusing on up to 15 blue-chip companies. Key indicators like MACD, RSI, and Bollinger Bands offer clear buy, sell, or hold recommendations. It actively monitors the Philippine Stock Exchange, notifying users of significant deviations. Developed in Python and C#, both models were trained on a consistent 5-year dataset for reliable performance. Results show high prediction accuracy, averaging 98.923% (high price) and 98.741% (low price), benefiting investment strategies. The study compared SPOT-allocated portfolios to random ones, revealing a notable difference in profit/loss allocation. This underscores its value for informed and profitable decision-making in stock trading.

关键词：

Properties of the Geometry of Solutions and Capacity of Multilayer Neural Networks with Rectified Linear Unit Activations

学校读者我要写书评

暂无评论

Physical Review Letters 2019年第17期123卷 170602-170602页

作者： Carlo Baldassi Enrico M. Malatesta Riccardo Zecchina Artificial Intelligence Lab Institute for Data Science and Analytics Bocconi University Milano 20135 Italy

Rectified linear units (ReLUs) have become the main model for the neural units in current deep learning systems. This choice was originally suggested as a way to compensate for the so-called vanishing gradient problem which can undercut stochastic gradient descent learning in networks composed of multiple layers. Here we provide analytical results on the effects of ReLUs on the capacity and on the geometrical landscape of the solution space in two-layer neural networks with either binary or real-valued weights. We study the problem of storing an extensive number of random patterns and find that, quite unexpectedly, the capacity of the network remains finite as the number of neurons in the hidden layer increases, at odds with the case of threshold units in which the capacity diverges. Possibly more important, a large deviation approach allows us to find that the geometrical landscape of the solution space has a peculiar structure: While the majority of solutions are close in distance but still isolated, there exist rare regions of solutions which are much more dense than the similar ones in the case of threshold units. These solutions are robust to perturbations of the weights and can tolerate large perturbations of the inputs. The analytical results are corroborated by numerical findings.

关键词： artificial neural networks Disordered systems

Semantics-Guided Contrastive Network for Zero-Shot Object Detection

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Yan, Caixia Chang, Xiaojun Luo, Minnan Liu, Huan Zhang, Xiaoqin Zheng, Qinghua The Ministry of Education Key Lab of Intelligent Networks and Network Security National Engineering Lab for Big Data Analytics School of Electronic and Information Engineering Xi'an Jiaotong University Shaanxi710049 China The ReLER Lab AAAI Faculty of Engineering and Information Technology University of Technology Sydney Australia The College of Computer Science and Artificial Intelligence Wenzhou University China

Zero-shot object detection (ZSD), the task that extends conventional detection models to detecting objects from unseen categories, has emerged as a new challenge in computer vision. Most existing approaches tackle the ZSD task with a strict mapping-transfer strategy that may lead to suboptimal ZSD results: 1) the learning process of these models neglects the available semantic information on unseen classes, which can easily bias towards the seen categories;2) the original visual feature space is not well-structured for the ZSD task due to the lack of discriminative information. To address these issues, we develop a novel Semantics-Guided Contrastive Network for ZSD, named ContrastZSD, a detection framework that first brings contrastive learning mechanism into the realm of zero-shot detection. Particularly, ContrastZSD incorporates two semantics-guided contrastive learning subnets that contrast between region-category and region-region pairs respectively. The pairwise contrastive tasks take advantage of supervision signals derived from both the ground truth label and class similarity information. By performing supervised contrastive learning over those explicit semantic supervision, the model can learn more knowledge about unseen categories to avoid the bias problem to seen concepts, while optimizing the visual data structure to be more discriminative for better visual-semantic alignment. Extensive experiments are conducted on two popular benchmarks for ZSD, i.e., PASCAL VOC and MS COCO. Results show that our method outperforms the previous state-of-the-art on both ZSD and generalized ZSD tasks. © 2021, CC BY.

关键词： Object detection

ICDAR 2021 competition on on-line signature verification

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Tolosana, Ruben Vera-Rodriguez, Ruben Gonzalez-Garcia, Carlos Fierrez, Julian Rengifo, Santiago Morales, Aythami Ortega-Garcia, Javier Ruiz-Garcia, Juan Carlos Romero-Tapiador, Sergio Jiang, Jiajia Lai, Songxuan Jin, Lianwen Zhu, Yecheng Galbally, Javier Diaz, Moises Ferrer, Miguel Angel Gomez-Barrero, Marta Hodashinsky, Ilya Sarin, Konstantin Slezkin, Artem Bardamova, Marina Svetlakov, Mikhail Saleem, Mohammad Szücs, Cintia Lia Kovari, Bence Pulsmeyer, Falk Wehbi, Mohamad Zanca, Dario Ahmad, Sumaiya Mishra, Sarthak Jabin, Suraiya Biometrics and Data Pattern Analytics Lab UAM Spain South China University of Technology China Guangdong Artificial Intelligence and Digital Economy Laboratory China European Commission Joint Research Centre Italy Universidad del Atlantico Medio Spain Universidad de las Palmas de Gran Canaria Spain Hochschule Ansbach Germany Tomsk State University of Control Systems and Radioelectronics Russia Budapest University of Technology and Economics Hungary Machine Learning and Data Analytics Lab FAU Germany Jamia Millia Islamia India

This paper describes the experimental framework and results of the ICDAR 2021 Competition on On-Line Signature Verification (SVC 2021). The goal of SVC 2021 is to evaluate the limits of on-line signature verification systems on popular scenarios (office/mobile) and writing inputs (stylus/finger) through large-scale public databases. Three different tasks are considered in the competition, simulating realistic scenarios as both random and skilled forgeries are simultaneously considered on each task. The results obtained in SVC 2021 prove the high potential of deep learning methods. In particular, the best on-line signature verification system of SVC 2021 obtained Equal Error Rate (EER) values of 3.33% (Task 1), 7.41% (Task 2), and 6.04% (Task 3). SVC 2021 will be established as an on-going competition12, where researchers can easily benchmark their systems against the state of the art in an open common platform using large-scale public databases such as DeepSignDB13 and SVC2021_EvalDB14, and standard experimental protocols. © 2021, CC BY-NC-ND.

关键词： Deep learning

Properties of the geometry of solutions and capacity of multi-layer neural networks with Rectified Linear Units activations

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Baldassi, Carlo Malatesta, Enrico M. Zecchina, Riccardo Artificial Intelligence Lab Institute for Data Science and Analytics Bocconi University Milano Italy

Rectified Linear Units (ReLU) have become the main model for the neural units in current deep learning systems. This choice has been originally suggested as a way to compensate for the so called vanishing gradient problem which can undercut stochastic gradient descent (SGD) learning in networks composed of multiple layers. Here we provide analytical results on the effects of ReLUs on the capacity and on the geometrical landscape of the solution space in two-layer neural networks with either binary or real-valued weights. We study the problem of storing an extensive number of random patterns and find that, quite unexpectedly, the capacity of the network remains finite as the number of neurons in the hidden layer increases, at odds with the case of threshold units in which the capacity diverges. Possibly more important, a large deviation approach allows us to find that the geometrical landscape of the solution space has a peculiar structure: While the majority of solutions are close in distance but still isolated, there exist rare regions of solutions which are much more dense than the similar ones in the case of threshold units. These solutions are robust to perturbations of the weights and can tolerate large perturbations of the inputs. The analytical results are corroborated by numerical findings. Copyright © 2019, The Authors. All rights reserved.

关键词： Multilayer neural networks