检索结果-内蒙古大学图书馆

37th Conference on Neural Information processing systems (NeurIPS)

作者： Chen, Xiangning Liang, Chen Huang, Da Real, Esteban Wang, Kaiyuan Pham, Hieu Dong, Xuanyi Luong, Thang Hsieh, Cho-Jui Lu, Yifeng Le, Quoc v. Google Mountain View CA 94043 USA UCLA Los Angeles CA 90095 USA

ISBN: (纸本)9781713899921

We present a method to formulate algorithm discovery as program search, and apply it to discover optimization algorithms for deep neural network training. We leverage efficient search techniques to explore an infinite and sparse program space. To bridge the large generalization gap between proxy and target tasks, we also introduce program selection and simplification strategies. Our method discovers a simple and effective optimization algorithm, Lion (EvoLved Sign Momentum). It is more memory-efficient than Adam as it only keeps track of the momentum. Different from adaptive optimizers, its update has the same magnitude for each parameter calculated through the sign operation. We compare Lion with widely used optimizers, such as Adam and Adafactor, for training a variety of models on different tasks. On image classification, Lion boosts the accuracy of viT by up to 2% on imageNet and saves up to 5x the pre-training compute on JFT. On vision-language contrastive learning, we achieve 88.3% zero-shot and 91.1% fine-tuning accuracy on imageNet, surpassing the previous best results by 2% and 0.1%, respectively. On diffusion models, Lion outperforms Adam by achieving a better FID score and reducing the training compute by up to 2.3x. For autoregressive, masked language modeling, and fine-tuning, Lion exhibits a similar or better performance compared to Adam. Our analysis of Lion reveals that its performance gain grows with the training batch size. It also requires a smaller learning rate than Adam due to the larger norm of the update produced by the sign function. Additionally, we examine the limitations of Lion and identify scenarios where its improvements are small or not statistically significant.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Smart Healthcare: A Logical Development of Medical image processing Based Sleep Apnea Disease Detection Using Elevated Neural Classification Principle

Smart Healthcare: A Logical Development of Medical Image Pro...

引用

2024 International Conference on Innovative Computing, Intelligent Communication and Smart Electrical systems, ICSES 2024

作者： Jagadeesh, P. Radhamani, v. Ravi, T. Charulatha, G. Shahare, D.Y. Aljawawdeh, Hamzeh J. Vel Tech Multi Tech Dr. Rangarajan Dr. Sakunthala Engineering college Department of Electronics and Communication Engineering Avadi Chennai India Rajalakshmi Engineering College Department of Ece Chennai India Sathyabama Institute of Science and Technology Department of Ece Chennai India Selvam College of Technology Department of Biomedical Engineering Namakkal India Yeshwantrao Chavan College of Engineering Department of Mechanical Engineering Nagpur India Zarqa University Faculty of information technology Department of Software Engineering Zarqa Jordan University of Business and Technology Jeddah21448 Saudi Arabia

ISBN: (纸本)9798331543617

Entirely common sleep disorder is the dangerous wakefulness (Apnea) of breathing during snoring. In this study, we explore the possibility of how medical image processing can be useful and accurate method for sleep disordered breathing detection. The new approach named as Sleep Apnea Analyzing Strategy (SAAS) is firstly tested with the help of existing method namely image base Sleep Apnea Analyzer (ISAA) that use less number or data to be processed and hence running on portable device delivering results much faster than other available methods, both being cross-validated against each other. Sleep Apnea image-based Detector using Ephysiology - A new Algorithm (SAAS) We rely on sophisticated image processing techniques to extract relevant characteristics from medical images (such as x-rays, CT scans or MRIs) which are indicative of patterns typically associated with sleep apnea. The proposed methodology includes image preprocessing to improve the quality of images, feature extraction for quantification and characterization of all relevant information during OSA trials in GUI interface through well-known classification algorithms to classify patients with at least 20 de-saturations without sleep apnea. Our research aims to help in the creation of automatic diagnostic methods that will lead to early detection and treatment for sleep apnea. Our objective with this work is to provide a fair evaluation of the proposed technique in terms of sensitivity, specificity and accuracy through using a large dataset. © 2024 IEEE.

关键词： Sleep research

来源：评论

学校读者我要写书评

暂无评论

Log Grading and Knot Identification by Oblique X-Ray Scanning

引用

JOURNAL OF NONDESTRUCTIvE EvALUATION, DIAGNOSTICS AND PROGNOSTICS OF ENGINEERING systems 2023年第2期6卷 021002页

作者： Omori, C. S. Schajer, G. S. FPInnovations Vancouver BC V6T 1Z4 Canada Univ British Columbia Dept Mech Engn Vancouver BC V6T 1Z4 Canada

The presence and location of knots within cut lumber substantially controls the physical properties and commercial value of the material. Thus, there is great practical interest in developing ways of choosing the cutting pattern for a log in a sawmill to optimize the arrangement of knots in the resulting cut lumber. X-rays can image the interior of a log to detect the arrangement of the knots;however, traditional radiography measurements are two-dimensional in character and cannot provide the needed depth information. Conversely, computed tomography (CT) can provide the required spatial details but is challenging to do in practice because of its complexity and cost. The research here aims to overcome these concerns by employing a novel "oblique" scanning technique that uses radiography to determine knot orientations with both reasonable accuracy and low cost. image processing and detection algorithms are developed to locate and orientate the knots automatically within the scanned logs. Detection metrics of precision and recall are used to analyze the performance of the detection algorithm. Results indicate that the oblique scanning method is a viable way to detect and orientate knots within logs with both reasonable accuracy and low cost compared to existing methods. In initial tests, an average circumferential angle accuracy within 15 deg was achieved, with the detection algorithm being able to detect between 60% and 80% of the knots present within the log.

关键词： lumber X-ray knots feature detection diagnostic feature extraction imaging product quality control radiography testing methodologies

来源：评论

学校读者我要写书评

暂无评论

v-slam-Aided photogrammetry to process fisheye multi-camera systems sequences 10

V-slam-Aided photogrammetry to process fisheye multi-camera ...

引用

10th International Workshop 3D-ARCH "3D virtual Reconstruction and visualization of Complex Architectures"

作者： Elalailyi, A. Perfetti, L. Fassi, F. Remondino, F. Italy Universita Degli Studi di Brescia Italy Trento Italy

The advent of mobile mapping systems (MMSs) and computer vision algorithms has enriched a wide range of navigation and mapping tasks such as localisation, 3D motion estimation and 3D mapping. This study focuses on visual Simultaneous Localisation and Mapping (v-SLAM) in the context of two in-houses MMSs: Ant3D, a patented five-fisheye multi-camera rig and GeoRizon, a highresolution stereo fisheye rig. The aim is to leverage v-SLAM to enhance the systems performance in near-real-Time and non-real-Time 3D reconstruction applications. The research investigates both Monocular and Stereo v-SLAM applied to both MMSs and tackles the challenge of combining the v-SLAM estimated trajectory of one or a pair of cameras with known multi-camera relative orientation. We propose a state-of-The-Art code that serves as a flexible and extensible platform for MMSs image acquisition and processing, along with an adapted version of the well-established ORB-SLAM3.0. Evaluation is performed in a cultural heritage challenging setup: The Minguzzi spiral staircase in the Duomo di Milano Cathedral. Performed tests highlight that introducing v-SLAM trajectories as well as pre-calibrated interior orientation and multi-camera constraints improve speed, applicability and accuracy of 3D surveys. © 2024 International Society for Photogrammetry and Remote Sensing. All rights reserved.

关键词： Motion estimation

来源：评论

学校读者我要写书评

暂无评论

image Based Security System in cloud resource Scheduling 5

Image Based Security System in cloud resource Scheduling

引用

5th International Conference on Information Management and Machine Intelligence, ICIMMI 2023

作者： Chakraborty, Koushik Dora Pravina, C.T. Arya, Nancy Sharma, Sunil Rakesh, Shanu Kuttan Singh, Arvind Kunar Adamas University Kolkata India Department of Mathematics Vel Tech Rangarajan Dr Sagunthala R & D Institute of Science and Technology Avadi India Department of Computer Science and Engineering Shree Guru Gobind Singh Tricentenary University Gurugram India Department of Computer Science and Engineering Vivekananda Global University Jaipur India Department of Computer Science and Engineering Chouksey Engineering College Bilaspur Chhattisgarh India Department of Computer Science and Engineering S.V.N.I.E.T Barabanki India

ISBN: (纸本)9798400709418

The popularity of cloud computing in scheduling work processes, particularly logical work processes, is growing. The cloud computing environment may encounter significant issues with execution time and cost during resource allocation, resulting in disruptions in customer support quality. Text Passwords are right now a totally utilized insistence part, however graphical passwords and biometrics are likewise utilized. However, we really want to improve both comfort and security in any system to make it extremely secure. image-based security systems leverage visual information and image processing techniques to enhance security measures. These systems use images or video streams from cameras to monitor, detect, and respond to security threats. Most likely, the framework employs our algorithm to improve performance. It utilizes parts of both Graphical and Text Secret phrase plans. Advanced image processing algorithms analyze captured images for various security-related purposes, such as object detection, facial recognition, and anomaly detection. In contrast to the most well-known scheduling algorithms, such as multi-objective scheduling (MOS) and cost-conscious scheduling, the various scheduling algorithms in this proposed hybrid algorithm mannequin are used to make cloud services secure and highly available to end users. Identifying and verifying individuals based on facial features. This is used for access control, identity verification, and tracking individuals of interest. Analyzing images to detect unauthorized persons or objects in restricted areas, triggering alerts or alarms. In this investigation, we attempt to demonstrate the execution of various algorithm systems for the most well-known three static booking calculations. All the calculations are working out the all-out execution cost and timing of various calculations. © 2023 ACM.

关键词： Security systems

来源：评论

学校读者我要写书评

暂无评论

Shadow Removal in Off-Road Terrain Perception with Multi-Sensor Signal processing

Shadow Removal in Off-Road Terrain Perception with Multi-Sen...

引用

2024 NDIA Michigan Chapter Ground vehicle systems Engineering and Technology Symposium, GvSETS 2024

作者： Gardner, S.D. Hoxie, D. Bowen, N. Misko, S. Haider, M.R. Smereka, J. Jayakumar, P. vantsevich, v. United States Electrical and Computer Engineering Department United States US Army Combat Capabilities Development Command Ground Vehic United States

Autonomous vehicle navigation requires signal processing of the vehicle's sensors to provide meaningful information to the planners such that challenging artifacts like shadows, rare events, obstructive vegetation, etc. are identified properly, avoiding ill-informed navigation. Using a single algorithm such as semantic segmentation of camera images is often not enough to identify those challenging features but can be overcome by processing more than one type of sensor and fusing their results. In this work, semantic segmentation of camera image and LiDAR point cloud signals is performed using Echo State Networks to overcome the challenge of shadows identified as obstructions in off-road terrains. The coordination of algorithms processing multiple sensor signals is shown to avoid unnecessary road obstructions caused by high-contrast shadows for more informed navigational planning. © Rights reserved by the National Defense Industrial Association (NDIA) Michigan Chapter, authors and their respected organizations.

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

Severity Based Detection of Conjunctivitis and Drug Recommendation System Using CNN 12

Severity Based Detection of Conjunctivitis and Drug Recommen...

引用

12th IEEE International Conference on Communication systems and Network Technologies, CSNT 2023

作者： Prakash, v.R. Sasikumar, S. Hari, M. Alavala, venkataRamiReddy Hindustan Institute of Technology and Science Department of Electronics & Communication Engineering Chennai India

ISBN: (纸本)9781665462617

The classification of an eye disease may be useful in the clinical setting for describing the current state of the eye, evaluating the effectiveness of treatment, and selecting the best course of action. Every classification scheme should satisfy the following requirements: clinical character (i.e., be performable by any physician equipped with a standard non-invasive method), repeatability, and meaningfulness (i.e., objective and of clinical relevance to the patient). The image processing approach is appropriate for purely descriptive disease detection. When evaluating the efficacy of a treatment, it is not particularly useful. The easiest way to communicate that result is probably by comparing disease types and stages. To diagnose just one disease, many machine learning algorithms were utilised in the past. Convolutional neural networks (CNNs), for example, Deep learning systems, are capable of inferring a hierarchical depiction of pictures in order to differentiate between proper and improper ocular patterns for diagnostic reasons. In order to assist the doctor (as an aid to doctor) to diagnose viral conjunctivitis at a faster and efficient way based on the severity, we propose our work to use Multi-dimensional (Md-CNN) architecture of conv-net for classification of medical eye image processing which has a higher efficiency in-case of medical image processing and to provide suggestions on medication based on the severity of the infection. © 2023 IEEE.

关键词： Diagnosis

来源：评论

学校读者我要写书评

暂无评论

An Advanced Feature Extraction Approach for Classifying and Categorizing Complex Data by Using the MMRDL Framework

An Advanced Feature Extraction Approach for Classifying and ...

引用

2023 International Conference on Self Sustainable Artificial Intelligence systems, ICSSAS 2023

作者： Krishna, I.M.v. Devi, T. Uma GITAM Deemed to Be University Dept. of Computer Science India

ISBN: (纸本)9798350300857

Nowadays, Machine learning (ML) plays a significant role in processing multimedia based on user search queries. To develop an efficient classification model that processes the high dimensional data, an ensemble model combined with deep feature extraction and robust classifier. Combining the three algorithms generate the stochastic belief space in this work. The planning of these algorithms based on maximum entropy reinforcement learning (ME-RL), and generative adversary model (GAM). The comparison between various unmodeled adversarial methods is suggested to achieve robustness because the malicious behaviors were applied in the simulation framework to reduce the agent's action predictability. Guided Deep Learning (DL), the reinforcement-based approach, can be employed for multi-model categorization. The text and image data applied on a single neural network method may categorize. The proposed system properly understands the belief space policy using feature extraction on text and image datasets. The proposed model shows the maximum entropy depends on the input dataset. © 2023 IEEE.

关键词： Feature extraction

来源：评论

学校读者我要写书评

暂无评论

CIFAR-10 Dataset image Classification Using CNN 8th

CIFAR-10 Dataset Image Classification Using CNN

引用

8th International Conference on Information System Design and Intelligent Applications, ISDIA 2024

作者： Sucharitha, M. Sesha Talpa Sai, P.H.v. vijayagovindan, Sanjith Razmin, Fabina Ravindran, Byju Kiran Babu, vadapalli v. S. Naveen Kumar, G.S. Bhaumik, Amiya School of Electronics Engineering VIT-AP University Amaravati India Malla Reddy College of Engineering and Technology Hyderabad India School of Information Technology Lincoln University College Petaling Jaya Malaysia Malla Reddy University Hyderabad India Lincoln University College Petaling Jaya Malaysia

ISBN: (纸本)9789819786657

image classification is an essential technology that is widely used in all aspects of human life. This work combined multiple image feature sources using deep learning algorithms to identify photos from the publicly available CIFAR10 image dataset. The same architecture, which consists of convolution modification and max-pooling of layers connected with several completely linked layers, is used to build most conventional convolutional neural networks (CNNs). This system consists of two steps: Pre-processing and Classification. The modified bilateral filtering method was employed in this study’s preprocessing steps. This study used convolutional neural network (CNN) architectures as suggested by the vGG-19 model to classify the CIFAR10 image dataset. The accuracy, precision, recall, and F1-score of the results are evaluated after simulating the suggested model. The testing results demonstrate that the proposed method reach a recall value of 96.4%, F1-score of 94.1%, accuracy of 97.3%, and maximum precision of 96.1%. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Wiener filtering

来源：评论

学校读者我要写书评

暂无评论

Skin Cancer Detection Using CNN and Ensemble Learning 1

Skin Cancer Detection Using CNN and Ensemble Learning

引用

1st IEEE International Conference on Advances in Signal processing, Power, Communication, and Computing, ASPCC 2024

作者： Samantaray, Ruturaj Afrin, Karishma Behera, Bichitrananda Bag, Kirtimayee Tata Consultancy Services Limited Tcs BaNCS Bangalore560066 India C. V. Raman Global University Dept. of Computer Science and Engineering Bhubaneswar752054 India

ISBN: (纸本)9798350355093

The global prevalence of skin cancer has seen a concerning increase over recent decades, now accounting for one in every three cancer cases worldwide. The unchecked growth of aberrant cells in the skin can lead to skin cancer, a potentially fatal illness. There are different types of skin cancer, namely, DF (Dermatofibroma), AKIEC (Actinic Keratoses and Intraepithelial Carcinoma), Mel (Melanoma), BCC (Basal Cell Carcinoma), vASC (vascular Lesions), BKL (Benign Keratosislike Lesions), and Nv (Melanocytic Nevi). Skin cancer can spread to nearby tissues in our body if it is not caught early. Presently, diagnosing skin cancer is a time-consuming and financially burdensome process for both patients and healthcare systems. Moreover, diagnosing skin cancer heavily relies on human cell detection. Skin cancers can be dangerous if they are not treated before they have a chance to spread. However, existing technology lacks the ability to detect skin cancer accurately. Traditional machine learning algorithms display limitations when classifying different classes of skin cancer. In this research, we have proposed skin cancer detection with classification using CNN and ensemble learning. CNN, known as the Convolutional Neural Network, represents a specialized architecture within deep learning algorithms primarily tailored for image recognition and feature extraction. It comprises convolutional, pooling, and fully connected layers in its mathematical structure. In this work, CNN (Convolutional Neural Network) is employed for image feature extraction, followed by utilizing three distinct classifier algorithms, namely, SvM (Support vector Machine), KNN (K-Nearest Neighbour), and DT (Decision Tree), integrated through an ensemble approach utilizing a voting system. Despite utilizing multiple classifiers, the model exhibits impressive performance, achieving a classification accuracy of 98%. Notably, our research leverages the HAM10000 dataset for this study. © 2024 IEEE.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：