Diabetic Retinopathy (DR) is one of the most severe sight-threatening disorders resulting from diabetes, and can eventually lead to blindness and visual impairment. Early detection and medical therapy can assist in co...
Diabetic Retinopathy (DR) is one of the most severe sight-threatening disorders resulting from diabetes, and can eventually lead to blindness and visual impairment. Early detection and medical therapy can assist in controlling and preventing DR progression. However, manual grading is extremely difficult and time-consuming because of the retina’s complicated structure. In this research, the Bat Optimization Algorithm- Refined Deep Residual Network (BOA-RDRN) is proposed for automatic diagnosis and grading of DR. Initially, the DIARETDB0 dataset is employed for the proposed method, and then pre-processing is performed for image denoising which eliminates noise. The Attention-based Fusion Network (AFU-Net) is used to segment the lesion region. The Grey-Level Co-occurrence Matrix (GLCM) and Local Ternary Pattern (LTP) are used to extract the features and BOA is employed to find an optimal subset of features in retinal images. Finally, the RDRN is utilized to classify the anomalies as normal or abnormal. The BOA-RDRN achieves an accuracy of 99.42% compared to the existing methods such as Tunicate Swarm Spider Monkey optimization-based Refined Deep Residual Network (TSSMO-RDRN), Weighted Kernel Fuzzy C-Means Clustering and Dilation-Based Function (WKFM-DBF), Improved Harris Hawk Optimization-based Convolution Neural Network (IHHO-CNN), and Enhanced Fuzzy C-Means Clustering-Resnet 152 respectively.
The accurate profile of learners on the online platform based on data analysis refers to the comparison of users with certain behavioral characteristics and existing in the database and not discovered, and then determ...
详细信息
The accurate profile of learners on the online platform based on data analysis refers to the comparison of users with certain behavioral characteristics and existing in the database and not discovered, and then determine the potential interests and concerns of the group based on this information, and recommend specific groups or products for them to meet the needs of users. Accurate profiling based on data analysis refers to the comparison of users with certain behavioral characteristics that do not exist or have not been discovered in the database to determine the potential interests and concerns of this group, which can provide a basis for recommending specific groups of people and products. First of all, this paper studies the data analysis technology, mainly expounds the data preprocessing, data visualization, behavior image processing and xAPI research, of which xAPI is a standard for collecting and storing data, mainly referring to data preprocessing and mining, statistical analysis, etc. Secondly, the characteristic analysis of the online learning platform is carried out, and finally the accurate portrait of online learners based on xAPI is constructed.
In recent years, a number of algorithms for unattended luggage detection have been proposed. They use both classical computer vision and neural network approaches. Classical methods do not work well on crowded scenes ...
详细信息
In recent years, a number of algorithms for unattended luggage detection have been proposed. They use both classical computer vision and neural network approaches. Classical methods do not work well on crowded scenes and when long time tracking is required. Neural networks face the fact that unattended objects are very diverse, that requires large representative training dataset and heavy network architecture. In this paper, we propose the Network Output Background Subtraction (NOBS): a real-time algorithm based on the use of a lightweight neural network trained on general purpose dataset with no use of any specialized dataset of abandoned luggage. It allows you to use the advantages of the neural network approach (the ability to track objects more accurately and for a long time) and partially get rid of its disadvantages: the need for an extensive training data set that reflects the large variability of detected objects, and high resource requirements.
The concept of near-memory computing (NMC) has emerged as a promising solution to address the memory wall challenges faced by future computing architectures. By utilizing modern systems that integrate 3D-stacked DRAM ...
The concept of near-memory computing (NMC) has emerged as a promising solution to address the memory wall challenges faced by future computing architectures. By utilizing modern systems that integrate 3D-stacked DRAM memory, the NMC paradigm minimizes unnecessary data movement between the memory subsystem and the CPU. FPGA vendors have incorporated 3D-stacked memories into their products to meet the increasing bandwidth requirements of memory-intensive applications, enabling FPGAs to compete with GPU solutions in terms of speed and energy efficiency. Recent NMC proposals focus on different dataprocessing workloads, including graph processing and machine learning. This work addresses the research questions of how to leverage the full bandwidth of 3D-stacked high-bandwidth memory and how to facilitate the adoption of the near-memory computing paradigm.
Handwritten digit recognition is a branch of machine learning in which a computer is taught to recognize hand-written numbers. Classification and regression are applied using deep learning and machine learning algorit...
Handwritten digit recognition is a branch of machine learning in which a computer is taught to recognize hand-written numbers. Classification and regression are applied using deep learning and machine learning algorithms. In this paper, we discussed the efficiency of different algorithms: Random Forest, KNN, Naive Bayes, SVM, CNN, and Decision Tree Algorithm. Those algorithms were applied in the handwritten digit recognition process using the MINIST data set. By comparing each algorithm by training every one of them on the data set, then by testing, we got the best algorithm to get the optimal results with the highest accuracy. After comparing the results of each algorithm, it has shown that the neural network has the best results with 97.3% classification accuracy followed by K- the nearest neighbor with 97.2%.
Concerning Deep Learning (esp. CNNs) workloads on integrated-GPU based heterogeneous programming models and mobile/embedded platforms, we propose a data-driven offline (batch) RL-oriented dynamic power management (DPM...
详细信息
Concerning Deep Learning (esp. CNNs) workloads on integrated-GPU based heterogeneous programming models and mobile/embedded platforms, we propose a data-driven offline (batch) RL-oriented dynamic power management (DPM) agent/governor design that uses an offline RL method to learn a simple, effective and robust policy from previously-collected static datasets without further interaction with the environment, concerning intrinsically expensive training costs of generic offpolicy online RL or/and imitation learning (IL). Our preliminary results show improvements up to 20.3% in terms of energy-delay product (EDP) and a reduced inference time of 2.2% on Tiny- YOLOv3/v4, compared to the default separate CPU and GPU governors.
The buzz surrounding artificial intelligence has made machine learning a hot topic right now. Although there are many tools for visualising data, scripting languages are primarily used for model training. The Orange D...
The buzz surrounding artificial intelligence has made machine learning a hot topic right now. Although there are many tools for visualising data, scripting languages are primarily used for model training. The Orange data Mining tool gives you a lot of ways to change how data is preprocessed, how it is displayed, how models are trained, and how models are tested. In order to establish which strategy has the best Classification Accuracy and Precision, the proposed research uses machine learning techniques to predict the size of an organization based on a variety of parameters, including employee experience, income, job type, employee type, etc. The effectiveness of various machine learning techniques, including Naive Bayes, random forests, support vector machines, neural network, logistic regression, was evaluated. Classification Accuracy evaluations are performed by cross validation. For this paper, we consulted the Kaggle dataset.
Traditional ceramic tile defect detection methods are often limited by the finite nature of feature representation and the influence of complex backgrounds, resulting in low model accuracy and generalization ability. ...
Traditional ceramic tile defect detection methods are often limited by the finite nature of feature representation and the influence of complex backgrounds, resulting in low model accuracy and generalization ability. This paper proposes a Shuffle Attention Mechanism-based YOLOv5 detection algorithm, aiming to enhance the accuracy and robustness of ceramic tile defect detection. It is compared and analyzed against the YOLOv5 algorithm, as well as improved YOLOv5 algorithms based on three different attention mechanisms: SimAM, GAM, and NAM. Experimental results demonstrate that this algorithm effectively captures crucial features related to defects in ceramic tile images, enabling more accurate detection of various types of ceramic tile defects and reducing false positives. In summary, this study demonstrates the effectiveness and potential of the Shuffle Attention YOLOv5 algorithm in ceramic tile defect detection.
Social Networks are integral part of our lives. Each of them has application programming interface (API) to access its data. People willingly store their private information, photos, thoughts and locations in numerous...
Social Networks are integral part of our lives. Each of them has application programming interface (API) to access its data. People willingly store their private information, photos, thoughts and locations in numerous forums. We decided to determine how much information we could obtain about a person using automatic cloud serverless architectures, social network’s APIs and advanced data Science models and algorithms. The paper shows how vulnerable privacy is and how easy it is to consolidate users’ information using modern cloud technologies. We were able to obtain geolocations, friends, similar profiles, predict influencers, and even predict missing friends using Machine Learning graph models. At the end of the day, people should always guard their private information and innovative social platform corporations should carefully think about what data could be given to third parties.
artificial intelligence (AI) is an emerging era that has proven to have a fantastic capacity to revolutionize how facts are captured and retrieved. AI-enabled systems offer an expansion of abilities to streamline the ...
artificial intelligence (AI) is an emerging era that has proven to have a fantastic capacity to revolutionize how facts are captured and retrieved. AI-enabled systems offer an expansion of abilities to streamline the system of accumulating, managing, storing, and retrieving data. AI-enabled structures provide several benefits, such as using predictive analytics and herbal Language processing (NLP) to extract records more effectively and correctly and automated pass-referencing competencies to simplify locating records. AI-enabled systems make indexing and saving statistics easier, allowing users to look for the relevant data they need quickly. AI-enabled structures offer higher security measures and encryption techniques for defensive touchy facts. Functions like facial recognition, gadget mastering algorithms, and get admission to hold data safe and far from malicious actors. Such systems may control and store statistics securely, correctly, and managed. AI-enabled systems provide an excellent suite of features for streamlining the manner of seizure and retrieval of statistics. AI answers, including predictive analytics, natural language processing, and automated cross-referencing, allow groups to quickly get the right of entry to the relevant information they need, even by providing higher safety features for included facts. In destiny, this technology might also grow to be even greater advanced, assisting businesses in gaining facts to electricity predictive analytics, helping in advertising efforts, and enhancing the organizational operation.
暂无评论