检索结果-内蒙古大学图书馆

International Seminar on Research of Information Technology and Intelligent Systems (ISRITI)

作者： Kus Andriadi Yaya Heryadi Lukas Wayan Suparta Ilvico Sonata Computer Science Department Binus Graduate Program Doctor of Computer Science Bina Nusantara University Jakarta Indonesia Cognitive Engineering Research Group (CERG) Universitas Katolik Indonesia Atma Jaya Jakarta Indonesia Department of Electrical Engineering Faculty of Industrial Technology Institut Teknologi Nasional Yogyakarta Yogyakarta Indonesia

Batik is an Indonesian world cultural heritage. Batik consists of many kinds of patterns depending on where the batik comes from, Batik-making techniques continue to develop along with technology development. Among the batik making techniques that are widely used are hand-written, stamping, and printing. Batik motifs have been widely used as research material, especially in the field of artificial intelligence. The diverse appearance of batik motifs has attracted many researchers to carry out research on making synthetic batik patterns, one of which uses a Generative Adversarial Network. This paper presents a synthetic batik pattern model based on the Wasserstein Generative Adversarial Network with Gradient Penalty. This model has been proven to create new synthetic batik patterns quite well and almost identical with images provided in the dataset, with the notes if the dataset provided is large.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Dielectric Bandpass Filter for Terahertz Communications

Dielectric Bandpass Filter for Terahertz Communications

引用

Asia-Pacific Conference on Microwave

作者： Linxi Chen Weijie Gao Jeffrey Hesler Christophe Fumeaux Withawat Withayachumnankul Terahertz Engineering Laboratory The University of Adelaide SA Australia Graduate School of Engineering Science Osaka University Toyonaka Osaka Japan Virginia Diodes Inc. Charlottesville VA USA School of Electrical Engineering and Computer Science The University of Queensland Australia QLD

ISBN: (数字)9798350363548

ISBN: (纸本)9798350363555

Toward realization of terahertz communications, we propose a method for designing terahertz bandpass filters on all-dielectric substrateless waveguide platforms. The proposed filter is constituted by one-dimensional photonic crystal as a low-pass filter and a two-dimensional photonic crystals as a high - pass filter. Leveraging cascaded photonic-crystal (PhC) claddings, wider stopbands are achieved to accommodate the requirements of different communications channels. The incorporation of both 2D and 1D PhC enables bandpass responses and improve design flexibility. With a low insertion loss of around 1.4 dB and a compact footprint, this novel family of designs is suitable for parallel channelized terahertz communications systems, effectively supporting the increase in the data rate.

关键词： Band-pass filters Optical losses Claddings Low-pass filters Photonic crystals Microwave theory and techniques Passband Microwave filters Substrates Terahertz communications

来源：评论

学校读者我要写书评

暂无评论

Interpretable Machine Learning for Multi-Class Crop Yield Prediction

Interpretable Machine Learning for Multi-Class Crop Yield Pr...

引用

Intelligent Cybernetics Technology & Applications (ICICyTA), International Conference on

作者： Aldian Nurcahyo Yaya Heryadi Lukas Wayan Suparta Ilvico Sonata Compuer Science Binus Graduate Program Bina Nusantara University Jakarta Indonesia Cognitive Engineering Research Group (CERG) Universitas Katolik Indonesia Atma Jaya Jakarta Indonesia Department of Electrical Engineering Faculty of Industrial Technology Institut Teknologi Nasional Yogyakarta Yogyakarta Indonesia Compuer Science Departement Binus Graduate Program Doctor of Computer Science Bina Nusantara University Jakarta Indonesia

Crop Yield Analysis and Prediction is a fast-expanding discipline that is critical for optimizing agricultural methods. A lack of trustworthy data is one of the challenges in estimating crop yields. We develop predictive models for 22 different fruits and vegetables data. The goals of this study are to create accurate and interpretable crop recommendation models. We used multiple machine learning (ML) models for multi-class crop production prediction to fulfill our research goal. We thoroughly examined the influence of climate and nutrient factors on crop yield, considering their complex interactions. To improve the dataset, augmented data techniques were applied. Configuring the parameters and fine-tuning the hyperparameters is our technique to increase the model performance. Furthermore, we employ explainable artificial intelligence (XAI) techniques and interpretability tools like Shapley Additive exPlanations (SHAP) to improve the interpretability of our prediction model. Our findings reveal that the XGBoost model has the best performance model with 99.86% accuracy, followed by SVM Poly Kernel with 99.32% and Random Forest with 98.82%. Feature selection and analysis are emphasized, particularly in regional agricultural contexts. This study contributes to the creation of accurate and interpretable crop recommendation models while also addressing the issue of untrustworthy data, providing useful insights for optimizing agricultural practices.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Quality Improvements in the electrical Design of Typical DC-DC Electronic Power Converter for High-Voltage Applications 8

Quality Improvements in the Electrical Design of Typical DC-...

引用

8th International Conference on engineering and Emerging Technologies, ICEET 2022

作者： Recto, King Harold A. Escoto, Miguel T. Hernandez, Noel B. Ateneo de Manila University Department of Electronics Computer and Communications Engineering Quezon City Philippines Electrical and Electronics Engineering Institute University of the Philippines Diliman Quezon City Philippines Centro Escolar University Post-doctoral Program in Quality Management of the Graduate School Manila Philippines

ISBN: (纸本)9781665491068

The green transition has brought about a worldwide-shift to the use of renewables as alternative energy sources. Because of this, high voltage DC has been a field of interest in power electronics due to its capability in efficiently transmitting power to reach remote areas. Typical DC-DC electronic power converter design has the capability of boosting voltage level to a certain extent but not enough for high-voltage applications. Using the DMAIC Approach of the Six Sigma, quality improvements in the design was proposed to stabilize operating frequency, bring up the output voltage from 100 VDC to 4,000 VDC and provide at least 10% margin of error in the apparent power measured at the output side to offset inherent losses in the electrical design. These proposed quality improvements were sustained by designing a control circuit that would ensure the attainment of the parameters set in the Define Phase of the DMAIC Approach. The result of the study is a practical 12 VDC to 4,000 VDC boost converter design that has incorporated all the proposed quality improvements. © 2022 IEEE.

关键词： Six sigma

来源：评论

学校读者我要写书评

暂无评论

UpCycling: Semi-supervised 3D Object Detection without Sharing Raw-level Unlabeled Scenes

UpCycling: Semi-supervised 3D Object Detection without Shari...

引用

International Conference on computer Vision (ICCV)

作者： Sunwook Hwang Youngseok Kim Seongwon Kim Saewoong Bahk Hyung-Sin Kim Department of Electrical and Computer Engineering Seoul National University SK Telecom Seoul Korea Graduate School of Data Science Seoul National University

Semi-supervised Learning (SSL) has received increasing attention in autonomous driving to reduce the enormous burden of 3D annotation. In this paper, we propose UpCycling, a novel SSL framework for 3D object detection with zero additional raw-level point cloud: learning from unlabeled de-identified intermediate features (i.e., "smashed" data) to preserve privacy. Since these intermediate features are naturally produced by the inference pipeline, no additional computation is required on autonomous vehicles. However, generating effective consistency loss for unlabeled feature-level scene turns out to be a critical challenge. The latest SSL frameworks for 3D object detection that enforce consistency regularization between different augmentations of an unlabeled raw-point scene become detrimental when applied to intermediate features. To solve the problem, we introduce a novel combination of hybrid pseudo labels and feature-level Ground Truth sampling (F-GT), which safely augments unlabeled multi-type 3D scene features and provides high-quality supervision. We implement UpCycling on two representative 3D object detection models: SECOND-IoU and PV-RCNN. Experiments on widely-used datasets (Waymo, KITTI, and Lyft) verify that UpCycling outperforms other augmentation methods applied at the feature level. In addition, while preserving privacy, UpCycling performs better or comparably to the state-of-the-art methods that utilize raw-level unlabeled data in both domain adaptation and partial-label scenarios.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Robust and Transferable Backdoor Attacks Against Deep Image Compression With Selective Frequency Prior

arXiv

引用

arXiv 2024年

作者： Yu, Yi Wang, Yufei Yang, Wenhan Guo, Lanqing Lu, Shijian Duan, Ling-Yu Tan, Yap-Peng Kot, Alex C. Lab Interdisciplinary Graduate Programme Nanyang Technological University Singapore Pengcheng Laboratory Shenzhen China School of Electrical and Electronic Engineering Nanyang Technological University Singapore School of Computer Science and Engineering Nanyang Technological University Singapore School of Computer Science Peking University Beijing China

Recent advancements in deep learning-based compression techniques have demonstrated remarkable performance surpassing traditional methods. Nevertheless, deep neural networks have been observed to be vulnerable to backdoor attacks, where an added pre-defined trigger pattern can induce the malicious behavior of the models. In this paper, we propose a novel approach to launch a backdoor attack with multiple triggers against learned image compression models. Drawing inspiration from the widely used discrete cosine transform (DCT) in existing compression codecs and standards, we propose a frequency-based trigger injection model that adds triggers in the DCT domain. In particular, we design several attack objectives that are adapted for a series of diverse scenarios, including: 1) attacking compression quality in terms of bit-rate and reconstruction quality;2) attacking task-driven measures, such as face recognition and semantic segmentation in downstream applications. To facilitate more efficient training, we develop a dynamic loss function that dynamically balances the impact of different loss terms with fewer hyper-parameters, which also results in more effective optimization of the attack objectives with improved performance. Furthermore, we consider several advanced scenarios. We evaluate the resistance of the proposed backdoor attack to the defensive pre-processing methods and then propose a two-stage training schedule along with the design of robust frequency selection to further improve resistance. To strengthen both the cross-model and cross-domain transferability on attacking downstream CV tasks, we propose to shift the classification boundary in the attack loss during training. Extensive experiments also demonstrate that by employing our trained trigger injection models and making slight modifications to the encoder parameters of the compression model, our proposed attack can successfully inject multiple backdoors accompanied by their corresponding triggers int

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

Dementia Detection by Shopping Refugees Support Robot 2

Dementia Detection by Shopping Refugees Support Robot

引用

2nd IEEE Global Conference on Life sciences and Technologies, LifeTech 2020

作者： Kishimoto, Yuta Arai, Kenichi Imai, Tetsuo Kobayashi, Toru Nagasaki University Computer and Information Science Program Graduate School of Engineering Nagasaki Japan Nagasaki University Division of Electrical Engineering and Computer Science Graduate School of Engineering Nagasaki Japan

ISBN: (纸本)9781728170633

We have developed the Shopping Refugees Support Robot that enables the elderly to order items through Social Networking Service (SNS) by voice conversation without relying on smartphones. This paper proposes a detection method of dementia signs using this Robot. Concretely, deterioration of cognitive function is detected by the evaluation of short-term memory utilizing order history of the Robot. We expect detection of dementia signs on a daily basis without making the elderly person conscious of having a dementia test, by information that only the person knows. We implemented a prototype and confirmed its functional validity. © 2020 IEEE.

关键词： Internet of things

来源：评论

学校读者我要写书评

暂无评论

Meta-optic accelerators for object classifier

Meta-optic accelerators for object classifier

引用

2023 Conference on Lasers and Electro-Optics, CLEO 2023

作者： Zheng, Hanyu Liu, Quan Zhou, You Kravchenko, Ivan I. Huo, Yuankai Valentine, Jason Department of Electrical and Computer Engineering Vanderbilt University NashvilleTN37212 United States Department of Computer Science Vanderbilt University NashvilleTN37212 United States Department of Mechanical Engineering Vanderbilt University NashvilleTN37212 United States Interdisciplinary Materials Science Program Vanderbilt University NashvilleTN37212 United States Center for Nanophase Materials Sciences Oak Ridge National Laboratory Oak RidgeTN37830 United States

A meta-optic platform for accelerating object classification is demonstrated. End-to-end design is used to co-optimize the optical and digital systems resulting in a high-speed and robust classifier with 93.1% accurac...

ISBN: (纸本)9781957171258

关键词：

来源：评论

学校读者我要写书评

暂无评论

A Design for Improvement of Visual SLAM in Dynamic Environments Using Feature-Point Removal of Moving Persons

A Design for Improvement of Visual SLAM in Dynamic Environme...

引用

International Automatic Control Conference

作者： Kai-Tai Song Ching-Hao Meng Institute of Electrical and Control Engineering College of Electrical and Computer Engineering National Yang Ming Chiao Tung University Hsinchu Taiwan Graduate Degree Program of Robotics College of Engineering National Yang Ming Chiao Tung University Hsinchu Taiwan

This paper presents a design to improve the robustness of visual SLAM(vSLAM). A processing step of feature-removal is added to the tracking thread of the conventional ORB-SLAM2 algorithm to improve the localization accuracy of a mobile robot in an environment with moving persons. Instance segmentation and motion tracking are intergrated to identify motion state of people in an image. ORB feature points belonging to moving persons are removed for further processing of the vSLAM pipeline. The advantage of this method is that the vSLAM can remove feature points of moving people, while retain those belonging to static people in the environment, which improves the accuracy of robot pose estimation. The improved ORB-SLAM2 algorithm has been implemented in a NVIDIA Xavier embedded system, which is integrated to a mobile robot. In practical robot navigation experiments, the average positioning error of the proposed method is within 4cm for 22.4m travel distance. Compared with conventional ORB-SLAM2, the average accuracy of our vSLAM method improves 97% in a dynamic environment with moving people.

关键词：

来源：评论

学校读者我要写书评

暂无评论

VeML: An End-to-End Machine Learning Lifecycle for Large-scale and High-dimensional Data

arXiv

引用

arXiv 2023年

作者： Le, Van-Duc Bui, Cuong-Tien Li, Wen-Syan Department of Electrical and Computer Engineering Seoul National University Korea Republic of Graduate School of Data Science Seoul National University Korea Republic of

An end-to-end machine learning (ML) lifecycle consists of many iterative processes, from data preparation and ML model design to model training and then deploying the trained model for inference. When building an end-to-end lifecycle for an ML problem, many ML pipelines must be designed and executed that produce a huge number of lifecycle versions. Therefore, this paper introduces VeML, a Version management system dedicated to end-to-end ML Lifecycle. Our system tackles several crucial problems that other systems have not solved. First, we address the high cost of building an ML lifecycle, especially for large-scale and high-dimensional dataset. We solve this problem by proposing to transfer the lifecycle of similar datasets managed in our system to the new training data. We design an algorithm based on the core set to compute similarity for large-scale, high-dimensional data efficiently. Another critical issue is the model accuracy degradation by the difference between training data and testing data during the ML lifetime, which leads to lifecycle rebuild. Our system helps to detect this mismatch without getting labeled data from testing data and rebuild the ML lifecycle for a new data version. To demonstrate our contributions, we conduct experiments on real-world, large-scale datasets of driving images and spatiotemporal sensor data and show promising results. © 2023, CC BY-NC-ND.

关键词： Machine learning

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：