检索结果-内蒙古大学图书馆

Proceedings of the 28th Pan-Hellenic Conference on Progress in Computing and informatics

作者： Nikitas N. Karanikolas Mateusz Zywczak Informatics and Computer Engineering University of West Attica Athens Greece

ISBN: (纸本)9798400713170

Usually, Stemmers are manually created by native target language speakers, on the basis of rules for suffix stripping or replacements. In such cases linguistic knowledge of the target language is a requirement. There are also approaches for automatic creation of stemmers. The later approaches are usually based on statistical processing of text corpuses to calculate the occurrences and other metrics for stems and affixes. In such cases linguistic knowledge of the target language is not needed. We have invented a semi-automatic methodology (and we have made an equivalent implementation) for Stemmer generation that neither linguistic knowledge of target language nor statistical processing of corpuses is needed. In our approach only Information Retrieval expertise is needed. Here we are evaluating our approach against the rule-based (linguistic-knowledge-based) approach. To do so we evaluate two stemmers by a native speaker of the target (Polish) language. The first Stemmer is rule-based (it incorporates linguistic knowledge of Polish language) and the second one is based on our methodology (only the experience of Information Retrieval Experts is incorporated). The results are interesting and hard to wait for, they are in favor of our methodology.

关键词： Automatic Stemmer Generation

来源：评论

学校读者我要写书评

暂无评论

Enhancing Smart Mobility Through Machine Learning and Neural Networks 7

Enhancing Smart Mobility Through Machine Learning and Neural...

引用

7th Edition Global Conference on Wireless and Optical Technologies, GCWOT 2024

作者： Nadeem, Muhammad Haider, Waleej Ahmed, Haris Khan, Sallar Sir Syed University of Engineering & Technology Department of Computer Science & Information Technology Karachi Pakistan School of Informatics & Cybersecurity Technological University Dublin Dublin Ireland

ISBN: (纸本)9798331534271

This study investigates the prediction of taxi trip durations in New York City using machine learning (ML) models and neural networks (NN). Three models Linear Regression, Random Forest Regressor, and a Neural Network were developed and evaluated on a dataset containing various trip details. The analysis highlighted the importance of geographic and temporal features in predicting trip durations. The Neural Network model outperformed others, achieving a mean squared error (MSE) of 12,577,211 and a root mean squared error (RMSE) of 3,546, significantly better than the comparative models. These results underscore its capacity to capture complex urban mobility patterns effectively. The superior performance of the Neural Network supports its potential to optimize traffic flow and reduce congestion in dense urban environments, offering substantial improvements to urban transportation systems. © 2024 IEEE.

关键词： Neural network models

来源：评论

学校读者我要写书评

暂无评论

Compact Yet Highly Accurate Printed Classifiers Using Sequential Support Vector Machine Circuits

arXiv

引用

arXiv 2025年

作者： Sertaridis, Ilias Besias, Spyridon Afentaki, Florentia Balaskas, Konstantinos Zervakis, Georgios Department of Computer Engineering & Informatics University of Patras Patras Greece

Printed Electronics (PE) technology has emerged as a promising alternative to silicon-based computing. It offers attractive properties such as on-demand ultra-low-cost fabrication, mechanical flexibility, and conformality. However, PE are governed by large feature sizes, prohibiting the realization of complex printed Machine Learning (ML) classifiers. Leveraging PE’s ultra-low non-recurring engineering and fabrication costs, designers can fully customize hardware to a specific ML model and dataset, significantly reducing circuit complexity. Despite significant advancements, state-of-the-art solutions achieve area efficiency at the expense of considerable accuracy loss. Our work mitigates this by designing area- and power-efficient printed ML classifiers with little to no accuracy degradation. Specifically, we introduce the first sequential Support Vector Machine (SVM) classifiers, exploiting the hardware efficiency of bespoke control and storage units and a single Multiply-Accumulate compute engine. Our SVMs yield on average 6x lower area and 4.6% higher accuracy compared to the printed state of the art. © 2025, CC BY.

关键词： Cost engineering

来源：评论

学校读者我要写书评

暂无评论

Late Breaking Results: Energy-Efficient Printed Machine Learning Classifiers with Sequential SVMs

arXiv

引用

arXiv 2025年

作者： Besias, Spyridon Sertaridis, Ilias Afentaki, Florentia Balaskas, Konstantinos Zervakis, Georgios Department of Computer Engineering & Informatics University of Patras Patras Greece

Printed Electronics (PE) provide a mechanically flexible and cost-effective solution for machine learning (ML) circuits, compared to silicon-based technologies. However, due to large feature sizes, printed classifiers are limited by high power, area, and energy overheads, which restricts the realization of battery-powered systems. In this work, we design sequential printed bespoke Support Vector Machine (SVM) circuits that adhere to the power constraints of existing printed batteries while minimizing energy consumption, thereby boosting battery life. Our results show 6.5x energy savings while maintaining higher accuracy compared to the state of the art. © 2025, CC BY.

关键词： Support vector machines

来源：评论

学校读者我要写书评

暂无评论

Exploring the Impact of Synthetic Data on Human Gesture Recognition Tasks Using GANs

arXiv

引用

arXiv 2024年

作者： Kontogiannis, George Tzamalis, Pantelis Nikoletseas, Sotiris Computer Engineering & Informatics University of Patras Patras Greece

In the evolving domain of Human Activity Recognition (HAR) using Internet of Things (IoT) devices, there is an emerging interest in employing Deep Generative Models (DGMs) to address data scarcity, enhance data quality, and improve classification metrics scores. Among these types of models, Generative Adversarial Networks (GANs) have arisen as a powerful tool for generating synthetic data that mimic real-world scenarios with high fidelity. However, Human Gesture Recognition (HGR), a subset of HAR, particularly in healthcare applications, using time series data such as allergic gestures, remains highly unexplored. In this paper, we examine and evaluate the performance of two GANs in the generation of synthetic gesture motion data that compose a part of an open-source benchmark dataset. The data is related to the disease identification domain and healthcare, specifically to allergic rhinitis. We also focus on these AI models’ performance in terms of fidelity, diversity, and privacy. Furthermore, we examine the scenario if the synthetic data can substitute real data, in training scenarios and how well models trained on synthetic data can be generalized for the allergic rhinitis gestures. In our work, these gestures are related to 6-axes accelerometer and gyroscope data, serving as multi-variate time series instances, and retrieved from smart wearable devices. To the best of our knowledge, this study is the first to explore the feasibility of synthesizing motion gestures for allergic rhinitis from wearable IoT device data using Generative Adversarial Networks (GANs) and testing their impact on the generalization of gesture recognition systems. It is worth noting that, even if our method has been applied to a specific category of gestures, it is designed to be generalized and can be deployed also to other motion data in the HGR domain. Copyright © 2024, The Authors. All rights reserved.

关键词： Generative adversarial networks

来源：评论

学校读者我要写书评

暂无评论

Real-Time Pattern Recognition in Augmented Reality Application Through the Integration of TensorFlow Lite and Unity

Real-Time Pattern Recognition in Augmented Reality Applicati...

引用

2024 International Conference on computer engineering, Network and Intelligent Multimedia, CENIM 2024

作者： Mulyo, Dwi Rante, Hestiasari Sukaridhoto, Sritrusta Arissabarno, Cahyo Politeknik Elektronika Negeri Surabaya Department of Creative Multimedia Technology Surabaya Indonesia Politeknik Elektronika Negeri Surabaya Department of Informatics and Computer Engineering Surabaya Indonesia

ISBN: (纸本)9798350368802

The integration of machine learning with augmented reality (AR) is vital for traditional batik preservation, especially in real-time pattern recognition. Traditional AR applications often rely on external servers, increasing latency and reducing interaction fluidity. This paper presents real-time batik pattern recognition using quantized TensorFlow Lite (TFLite) in a mobile AR application. Using the Single Shot MultiBox Detector (SSD) architecture with quantization, model size is reduced by 75%, while maintaining 96% accuracy, 0.06 seconds inference time, and 46 frames per second (FPS). Compared to YOLO, which achieves 93% accuracy and 40 FPS, TFLite offers faster inference and reduced computational load, making it ideal for mobile AR. By embedding TFLite into an AR platform, users can interact with batik patterns in real-time without server reliance. This research shows that quantized TFLite enhances mobile AR performance, balancing speed, accuracy, and efficiency. © 2024 IEEE.

关键词： Object detection

来源：评论

学校读者我要写书评

暂无评论

Covert Channel Detection Methods 20

Covert Channel Detection Methods

引用

20th IEEE International Conference on Emerging eLearning Technologies and Applications, ICETA 2022

作者： Ondov, Adrian Helebrandt, Pavol Institute of Computer Engineering and Applied Informatics Slovak University of Technology Bratislava Slovakia

ISBN: (数字)9798350320336

ISBN: (纸本)9798350320336

The modern networking world is being exposed to many risks more frequently every day. Most of systems strongly rely on remaining anonymous throughout the whole endpoint exploitation process. Covert channels represent risk since they ex-ploit legitimate communications and network protocols to evade typical filtering. This firewall avoidance sees covert channels frequently used for malicious communication of intruders with systems they compromised, and thus a real threat to network security. While there are commercial tools to safeguard computer networks, novel applications such as automotive connectivity and V2X present new challenges. This paper focuses on the analysis of the recent ways of using covert channels and detecting them, but also on the state-of-the-art possibilities of protection against them. We investigate observing the timing covert channels behavior simulated via injected ICMP traffic into standard network communications. Most importantly, we concentrate on enhancing firewall with detection and prevention of such attack built-in features. The main contribution of the paper is design for detection timing covert channel threats utilizing detection methods based on statistical analysis. These detection methods are combined and implemented in one program as a simple host-based intrusion detection system (HIDS). As a result, the proposed design can analyze and detect timing covert channels, with the addition of taking preventive measures to block any future attempts to breach the security of an end device. © 2022 IEEE.

关键词： Statistical methods

来源：评论

学校读者我要写书评

暂无评论

Implementation of Deep Learning on Obstacle Avoidance System PENSHIP Ship Robot 24

Implementation of Deep Learning on Obstacle Avoidance System...

引用

24th International Electronics Symposium, IES 2022

作者： Madani, Hendika Putra Wibowo, Iwan Kurnianto Bachtiar, Mochamad Mobed Politeknik Elektronika Negeri Surabaya Departement of Informatics and Computer Engineering Surabaya Indonesia

ISBN: (数字)9781665489713

ISBN: (纸本)9781665489713

This study proposes an autonomous system for detecting objects in the form of a ball which is assumed to be the trajectory of the PENSHIP ship robot. In following the trajectory, the ship robot is prohibited from being operated manually by remote control or using additional sensors such as ultrasonic sensors, laser sensors, proximity sensors or GPS tracking. This study uses an image/camera detection option using Deep Learning with YOLO model to identify obstacles in the ship robot's track area, which will be implemented on the PENSHIP ship robot. The resistance data will be labeled and use the YOLO model. The results of this object recognition system succeeded in detecting the type of obstacle in the form of a ball with a YOLOv5 model with an accuracy of 90% at a distance of 2-5 meters from the obstacle. In the process of forming the Obstacle Avoidance System, the centroid value was successfully detected for each object bounding box and the simulation test results from the Obstacle Avoidance System that classified the movement direction of the ship was successful with 100% accuracy with BLDC motor speed approximately 1 m/s and the distance between the object at 2-5 meters. With this system, the PENSHIP Robot Ship will continue to follow the trajectory and avoid crash with the obstacles in front of it. © 2022 IEEE.

关键词： Ships

来源：评论

学校读者我要写书评

暂无评论

Landsat Image Classification Based on K-Nearest Neighbor

Landsat Image Classification Based on K-Nearest Neighbor

引用

2022 International Seminar on Application for Technology of Information and Communication, iSemantic 2022

作者： Barito, Raynaldi Bismantaka Sanjaya, Muhammad Hafidh Arif, Fajar Muhammad Humam, Naufal Aji, Pri Nugroho Sari, Christy Atika Rachmawanto, Eko Hari Suprayogi Dian Nuswantoro University Informatics Engineering Faculty of Computer Science Semarang Indonesia

ISBN: (数字)9781665488396

ISBN: (纸本)9781665488396

Classification is the process of grouping classes and defining a class and determining the relationship between these classes. Landsat imagery with the distribution of residential areas and agricultural areas can be used to process information on the population density of a particular area. In this study, the classification process of residential images, factory images and rice fields images has been carried out with a total of 58 data. KNN was chosen as the classification algorithm considering the data used is quite simple and few. In this study, GLCM is used for feature extraction features, especially regarding image texture patterns. We have implemented values K=1 to k=11. The best accuracy value is obtained at k=1 which is 100%, while k=3, k=5 has obtained an accuracy of 96.15%. k=7 and k=9 can still be tolerated by getting 76.92% while at k=11 it only gets 57.69%. © 2022 IEEE.

关键词： Image texture

来源：评论

学校读者我要写书评

暂无评论

Optimization of Chasing and Catching Skills on Robot Soccer ERSOW 24

Optimization of Chasing and Catching Skills on Robot Soccer ...

引用

24th International Electronics Symposium, IES 2022

作者： Ihsan, Fadl Lul Hakim Dewantara, Bima Sena Bayu Wibowo, Iwan Kurnianto Politeknik Elektronika Negeri Surabaya Department of Informatics and Computer Engineering Surabaya Indonesia

ISBN: (数字)9781665489713

ISBN: (纸本)9781665489713

The wheeled soccer robot competition is held annually in Indonesia and internationally through the Indonesian Robot Contest and Robocup Middle Size League (MSL) competitions. The rate of occurrence of the ball moving freely on the field during the match is quite high. At this time the movement of the ERSOW robot to catch the ball is by chasing and following the path of the ball on the field without taking into account the speed of the ball. The movement of the ERSOW Robot with this method is very vulnerable and can still be optimized. The ball is very susceptible to failure to be caught due to being controlled by the opponent or the ball being out. In addition, chasing the ball requires a longer time due to the path of the robot that continues to change following the current position of the ball movement. So the development of robot skills in chasing and blocking the ball is needed to increase the chance of controlling the ball. Based on the results of the literature study, was found a ball blocking ability on the soccer ball intercept robot. This research is focused on how the robot can chase and catch the ball with the ability to intercept. The robot's ability to know the direction of the ball's movement and cut the ball's movement or intercept is needed. By utilizing data processing from vision to obtain ball speed data and speed algorithm calculations, a more optimal method of chasing and catching the ball is obtained. Based on the results of the 180-degree angle experiment, the success of ERSOW in catching the ball using this method was 94.7%. © 2022 IEEE.

关键词： Sports

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：