检索结果-内蒙古大学图书馆

Predicting secondary school student performance using a double particle swarm optimization-based categorical boosting model

引用

ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE 2023年第1期124卷

作者： Fan, Zongwen Gou, Jin Wang, Cheng Huaqiao Univ Coll Comp Sci & Technol Xiamen 361021 Peoples R China Huaqiao Univ Xiamen Key Lab Comp Vis & Pattern Recognit Xiamen 361000 Peoples R China

Knowing the potential students who will fail the final exam at early stages is very challenging but important for the decision-makers in the educational institutions to take proper actions to prevent them from failure. To accurately predict the secondary school student performance, we propose a double particle swarm optimization (PSO)-based categorical boosting (P2CatBoost) model based on the demographic, school period grades, and social/school related features. Considering the machine learning models are sensitive to their hyper-parameter settings, we introduce the PSO to optimize the fitness function. In addition, the threshold of a standard binary classification task is 0.5, which might not be the optimal value in real-world applications. Thus, we optimize this threshold by the PSO. To evaluate the performance of our proposed model, two datasets downloaded from the University of California, Irvine Repository and the Kaggle, respectively, are used. The experimental results showed that our proposed P2CatBoost has the best performance in terms of all the metrics used. Our proposed P2CatBoost has the best accuracy of 96.62% and 94.45% for the final grade prediction of the Mathematics and Portuguese courses, respectively. In addition, our proposed model outperforms the other models under comparison from 4.5% to 8.3%. The statistical analyses verify that our P2CatBoost can significantly outperform the comparing models. These results confirm the effectiveness of our double PSO for improving the performance of student performance prediction, indicating our proposed model could be a useful tool in educational institutions to improve the quality of education.

关键词： Double particle swarm optimization Educational data mining CatBoost Educational management

来源：评论

学校读者我要写书评

暂无评论

Organ-aware CBCT enhancement via dual path learning for prostate cancer treatment

引用

MEDICAL PHYSICS 2023年第11期50卷 6931-6942页

作者： Chen, Xu Pang, Yunkui Ahmad, Sahar Royce, Trevor Wang, Andrew Lian, Jun Yap, Pew-Thian Huaqiao Univ Coll Comp Sci & Technol Xiamen Peoples R China Huaqiao Univ Fujian Prov Univ Key Lab Comp Vis & Machine Learning Xiamen Peoples R China Huaqiao Univ Xiamen Key Lab Comp Vis & Pattern Recognit Xiamen Peoples R China Univ N Carolina Dept Comp Sci Chapel Hill NC USA Univ N Carolina Dept Radiol Chapel Hill NC USA Univ N Carolina Biomed Res Imaging Ctr Chapel Hill NC USA Univ N Carolina Dept Radiat Oncol Chapel Hill NC USA Univ N Carolina Dept Radiol Chapel Hill NC 27599 USA

BackgroundCone-beam computed tomography (CBCT) plays a crucial role in the intensity modulated radiotherapy (IMRT) of prostate cancer. However, poor image contrast and fuzzy organ boundaries pose challenges to precise targeting for dose delivery and plan reoptimization for adaptive *** this work, we aim to enhance pelvic CBCT images by translating them to high-quality CT images with a particular focus on the anatomical structures important for *** develop a novel dual-path learning framework, covering both global and local information, for organ-aware enhancement of the prostate, bladder and rectum. The global path learns coarse inter-modality translation at the image level. The local path learns organ-aware translation at the regional level. This dual-path learning architecture can serve as a plug-and-play module adaptable to other medical image-to-image translation *** evaluated the performance of the proposed method both quantitatively and qualitatively. The training dataset consists of unpaired 40 CBCT and 40 CT scans, the validation dataset consists of 5 paired CBCT-CT scans, and the testing dataset consists of 10 paired CBCT-CT scans. The peak signal-to-noise ratio (PSNR) between enhanced CBCT and reference CT images is 27.22 & PLUSMN;1.79, and the structural similarity (SSIM) between enhanced CBCT and the reference CT images is 0.71 & PLUSMN;0.03. We also compared our method with state-of-the-art image-to-image translation methods, where our method achieves the best performance. Moreover, the statistical analysis confirms that the improvements achieved by our method are statistically *** proposed method demonstrates its superiority in enhancing pelvic CBCT images, especially at the organ level, compared to relevant methods.

关键词： dual path learning image-to-image translation unsupervised image enhancement

来源：评论

学校读者我要写书评

暂无评论

Bridging the reality gap: A benchmark for physical reasoning in general world models with various physical phenomena beyond mechanics

引用

EXPERT SYSTEMS WITH APPLICATIONS 2025年 270卷

作者： Zhao, Pengyu Xu, Jinan Cheng, Ning Hu, Huiqi Zhang, Xue Xu, Xiuwen Jin, Zijian Meng, Fandong Zhou, Jie Han, Wenjuan Beijing Jiaotong Univ State Key Lab Adv Rail Autonomous Operat Beijing Peoples R China Beijing Jiaotong Univ Key Lab Big Data & Artificial Intelligence Transpo Minist Educ Beijing Peoples R China Beijing Jiaotong Univ Sch Comp Sci & Technol Beijing Peoples R China New York Univ New York NY USA Tencent Inc Pattern Recognit Ctr WeChat AI Beijing Peoples R China

While general world models have demonstrated excellent capability in modeling and simulating the world through video understanding and generation, their ability to reason about physical phenomena beyond mechanics remains underexplored. This includes crucial aspects like thermodynamics, electromagnetism, and optics, all of which are fundamental for simulating and predicting real-world dynamics. Existing benchmarks for evaluating physical reasoning in models often rely on datasets consisting solely of simulator-generated, virtual videos, limiting their generalizability to real-world scenarios. This limitation hinders the comprehensive evaluation of general world models' physical reasoning in real-world scenarios. To bridge this gap, we introduce a physical reasoning dataset constructed from real-world videos (named Physics-RW). Encompassing a broad spectrum of real-world phenomena-mechanics, thermodynamics, electromagnetism, and optics- Physics-RW offers a comprehensive evaluation platform. We evaluated the performance of general world models on the Physics-RW benchmark, and the experimental results indicate that these models exhibit limited physical reasoning capabilities, suggesting significant room for exploration in this area. Amore in-depth analysis of the experimental results was conducted, and several potential avenues for improvement were explored. Specifically, experiments involving virtual environment fine-tuning and physical knowledge injection via prompts demonstrated that the proposed avenues can effectively enhance zero-shot physical reasoning ability of general world models.1

关键词： Physical reasoning Benchmark General world models

来源：评论

学校读者我要写书评

暂无评论

Deep Spatial-Temporal Network Based on Residual Networks and Dilated Convolution for Traffic Flow Prediction 7

Deep Spatial-Temporal Network Based on Residual Networks and...

引用

IEEE 7th International Conference on Intelligent Transportation Engineering (ICITE)

作者： Dong, Liangming Zhang, Xinxin Liu, Lijuan Xiamen Univ Technol Coll Comp & Informat Engn Fujian Key Lab Pattern Recognit & Image Understan Xiamen Peoples R China

ISBN: (纸本)9781665460071

As modern cities become more intelligent, traffic flow forecasting is becoming increasingly significant in intelligent transport and urban governance. Accurate traffic forecasting can help the management and the residents to travel, and how to achieve accurate and efficient traffic forecasting is a problem that has been studied by many scholars. With the growth of artificial intelligence and the generation of large-scale traffic data, deep learning can be more effective than traditional probabilistic prediction models in traffic flow forecasting, but there are two more problems to be solved. Firstly, the use of too many levels of convolution often leads to over-fitting. Secondly, the state of traffic in a city can be impacted by external factors such as meteorological conditions, temperature, and wind speed. To solve this problem, we propose an end-to-end neural network model based on the residual network, dilated convolution, and call it STDCN. Convolution and dilated convolution are used in the model to obtain local spatial dependencies and distant spatial dependencies. We designed an external feature extractor to transform the external factors into a 1D tensor using the One-Hot Encoding.

关键词： traffic flow forecasts intelligent transport deep learning external factors dilated convolutional

来源：评论

学校读者我要写书评

暂无评论

DHLA: Dynamic Hybrid label Assignment for End-to-End Object Detection

引用

IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY 2025年第2期35卷 1055-1069页

作者： Hu, Zhiliang Chen, Si Hua, Yang Wang, Da-Han Zhu, Shunzhi Yan, Yan Xiamen Univ Technol Sch Comp & Informat Engn Fujian Key Lab Pattern Recognit & Image Understand Xiamen 361024 Peoples R China Xidian Univ State Key Lab Integrated Serv Networks Xian 710071 Peoples R China Queens Univ Belfast Inst Elect Commun & Informat Technol Sch Elect Elect Engn & Comp Sci Belfast BT7 1NN North Ireland Xiamen Univ Sch Informat Xiamen 361005 Peoples R China

The recent one-to-one label assignment plays a crucial role in removing the last non-differentiable component, i.e., Non-Maximum Suppression (NMS), used in the post-processing step of the one-to-many label assignment, thus building an efficient end-to-end detection system. However, due to the limited number of foreground samples, the one-to-one label assignment often suffers from insufficient representation learning, and its performance is inferior to that of traditional detectors trained using the one-to-many label assignment. To solve these problems, we introduce a novel Dynamic Hybrid label Assignment (DHLA) method, including a Hybrid Sample Selection (HSS) strategy and a Stage-aware Soft-label Adjustment (SSA) mechanism. In order to enhance the ability of representation learning of the one-to-one label assignment, the HSS strategy subtly integrates the one-to-many and the one-to-one label assignment rules to form a simple and effective hybrid assignment rule, where high-quality samples are selected for training according to an effective task consistency metric. Moreover, the SSA mechanism dynamically adjusts the contributions of different foreground samples at different training stages, thus effectively achieving the transition from one-to-many to one-to-one label assignment. In addition, we leverage a ranking loss function to widen the score gaps between the highest scoring position and surrounding areas for effectively removing duplicate bounding boxes. As a result, our method not only learns robust feature representations during training but also performs efficient end-to-end detection during inference. Extensive experiments demonstrate our method achieves competitive performance compared to state-of-the-art detectors on the challenging COCO and CrowdHuman datasets.

关键词： Training Detectors Object detection Feature extraction Circuits and systems Transformers Silicon Representation learning Measurement DH-HEMTs NMS-free label assignment dynamic soft label

来源：评论

学校读者我要写书评

暂无评论

Explainable Deep Learning System for Advanced Silicon and Silicon Carbide Electrical Wafer Defect Map Assessment

引用

IEEE ACCESS 2022年 10卷 99102-99128页

作者： Sarpietro, Riccardo Emanuele Pino, Carmelo Coffa, Salvatore Messina, Angelo Palazzo, Simone Battiato, Sebastiano Spampinato, Concetto Rundo, Francesco STMicroelectronics ADG Res & Dev Power & Discretes I-95121 Catania Italy Univ Catania Dept Elect Elect & Comp Engn Pattern Recognit & Comp Vis Lab PeRCe Lab I-95124 Catania Italy Univ Catania Image Proc Lab IP LAB Grp Dipartimento Matemat & Informat DMI I-95124 Catania Italy

The recent increasing demand of Silicon-on-Chip devices has triggered a significant impact on the industrial processes of leading semiconductor companies. The semiconductor industry is redesigning internal technology processes trying to optimize costs and production yield. To achieve this target a key role is played by the intelligent early wafer defects identification task. The Electrical Wafer Sorting (EWS) stage allows an efficient wafer defects analysis by processing the visual map associated to the wafer. The goal of this contribution is to provide an effective solution to perform automatic evaluation of the EWS defect maps. The proposed solution leverages recent approaches of deep learning both supervised and unsupervised to perform a robust EWS defect patterns classification in different device technologies including Silicon and Silicon Carbide. This method embeds an end-to-end pipeline for supervised EWS defect patterns classification including a hierarchical unsupervised system to assess novel defects in the production line. The implemented "Unsupervised Learning Block" embeds ad-hoc designed Dimensionality Reduction combined with Clustering and a Metrics-driven Classification Sub-Systems. The proposed "Supervised Learning Block" includes a Convolutional Neural Network trained to perform a supervised classification of the Wafer Defect Maps (WDMs). The proposed system has been evaluated on several datasets, showing effective performance in the classification of the defect patterns (average accuracy about 97%).

关键词： Production Deep learning Wavelength division multiplexing Convolutional neural networks Silicon carbide Feature extraction Artificial intelligence Clustering methods Artificial intelligence convolutional neural network explainable architectures hierarchical clustering

来源：评论

学校读者我要写书评

暂无评论

A UNIFIED FRAMEWORK FOR MULTI-INTENT SPOKEN LANGUAGE UNDERSTANDING WITH PROMPTING 49

A UNIFIED FRAMEWORK FOR MULTI-INTENT SPOKEN LANGUAGE UNDERST...

引用

49th IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Song, Feifan Huang, Lianzhe Wang, Houfeng Peking Univ Sch Comp Sci Beijing Peoples R China Peking Univ Natl Key Lab Multimedia Informat Proc Beijing Peoples R China Tencent Pattern Recognit Ctr WeChat AI Shenzhen Peoples R China

ISBN: (纸本)9798350344868;9798350344851

ChatGPT has demonstrated impressive capabilities in building conversations. However, for Spoken Language Understanding (SLU) with multiple intents, traditional approaches where Intent Detection and Slot Filling are jointly modeled with distinct formulations hinder networks from effectively extracting shared features. In this work, we describe a Prompt-based SLU (PromptSLU) framework, to intuitively unify two sub-tasks into the same form for a common pre-trained model. Specifically, variable intents are predicted first, then naturally embedded into prompts to guide slot-value inference from a semantic perspective. Furthermore, we are inspired by multi-task learning to introduce an auxiliary sub-task and a concise general objective, which helps to learn relationships among provided labels. Experiment results show that our framework outperforms several competitive baselines on two datasets. The source code is available at https://***/F2-Song/PromptSLU.

关键词： Spoken Language Understanding Multi-intent Classification Slot Filling Prompting

来源：评论

学校读者我要写书评

暂无评论

Iterative Semantic Transformer by Greedy Distillation for Community Question Answering

引用

IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING 2024年 32卷 3576-3588页

作者： Wu, Jinmeng Mu, Tingting Thiyagalingam, Jeyarajan Hong, Hanyu Hao, Yanbin Zhang, Tianxu Goulermas, John Y. Wuhan Inst Technol Dept Elect Engn & Elect Hubei Key Lab Opt Informat & Pattern Recognit Wuhan 430079 Peoples R China Univ Manchester Sch Comp Sci Manchester M13 9PL Lancs England Rutherford Appleton Lab Sci & Technol Facil Council Didcot OX11 0QX Oxon England Wuhan Inst Technol HuBei Key Lab Opt Informat & Pattern Recognit Wuhan 430079 Peoples R China Univ Sci & Technol China Sch Informat Sci & Technol Hefei 230026 Peoples R China Huazhong Univ Sci & Technol Wuhan 430074 Peoples R China Univ Liverpool Dept Comp Sci Liverpool L69 3BX Merseyside England

The semantic matching problem consists of recognizing if the candidate text is relevant to a particular input text. Semantic similarities can be determined from human-curated knowledge, but such knowledge may not be available in every language. Instead, statistical learning techniques have been applied, but these techniques circumvent the need for manual feature engineering by using large datasets to train models to perform semantic similarity scoring between portions of text or words. The pre-trained transformer provides a further mechanism to consolidate the information throughout a sentence into single sentence-level representations, but these representations may not be optimal for the matching task. As an alternative, we propose an interactive semantic transformer based on a greedy layer-wise framework to learn a distributed similarity representation for sentence pairs. The novelty of the architecture lies in an abstract representation of the semantic similarities created by three-stage learning strategies. Model training is accomplished through a greedy layer-wise training scheme, that incorporates both supervised and unsupervised learning. The proposed model is experimentally compared to state-of-the-art approaches on three different dataset types: the library TREC, the Yahoo!, and Stack Exchange community question datasets, and results show the proposed model outperforming other approaches.

关键词： Speech processing Semantic representation descriptive clustering feature selection pre-trained model greedy-layer wise sparse models Semantic representation descriptive clustering feature selection pre-trained model greedy-layer wise sparse models

来源：评论

学校读者我要写书评

暂无评论

Fed-FA: Theoretically Modeling Client Data Divergence for Federated Language Backdoor Defense 37

Fed-FA: Theoretically Modeling Client Data Divergence for Fe...

引用

37th Conference on Neural Information Processing Systems (NeurIPS)

作者： Zhang, Zhiyuan Chen, Deli Zhou, Hao Meng, Fandong Zhou, Jie Sun, Xu Peking Univ Sch Comp Sci Natl Key Lab Multimedia Informat Proc Beijing Peoples R China Tencent Inc WeChat AI Pattern Recognit Ctr Beijing Peoples R China

ISBN: (纸本)9781713899921

Federated learning algorithms enable neural network models to be trained across multiple decentralized edge devices without sharing private data. However, they are susceptible to backdoor attacks launched by malicious clients. Existing robust federated aggregation algorithms heuristically detect and exclude suspicious clients based on their parameter distances, but they are ineffective on Natural Language Processing (NLP) tasks. The main reason is that, although text backdoor patterns are obvious at the underlying dataset level, they are usually hidden at the parameter level, since injecting backdoors into texts with discrete feature space has less impact on the statistics of the model parameters. To settle this issue, we propose to identify backdoor clients by explicitly modeling the data divergence among clients in federated NLP systems. Through theoretical analysis, we derive the f-divergence indicator to estimate the client data divergence with aggregation updates and Hessians. Furthermore, we devise a dataset synthesization method with a Hessian reassignment mechanism guided by the diffusion theory to address the key challenge of inaccessible datasets in calculating clients' data Hessians. We then present the novel Federated F-Divergence-Based Aggregation (Fed-FA) algorithm, which leverages the f-divergence indicator to detect and discard suspicious clients. Extensive empirical results show that Fed-FA outperforms all the parameter distance-based methods in defending against backdoor attacks among various natural language backdoor attack scenarios.

关键词： Natural language processing systems

来源：评论

学校读者我要写书评

暂无评论

CRITID: enhancing CRITIC with advanced independence testing for robust multi-criteria decision-making

引用

SCIENTIFIC REPORTS 2024年第1期14卷 1-22页

作者： Zhang, Qiang Fan, Jiahui Gao, Chaobang Chengdu Univ Sch Comp Sci Chengdu 610106 Peoples R China Chengdu Univ Inst Higher Educ Sichuan Prov Key Lab Pattern Recognit & Intelligent Informat Pr Chengdu Peoples R China Jiangxi Normal Univ Sch Comp & Informat Engn Nanchang 330022 Peoples R China Chengdu Univ Sichuan Prov Dept Culture & Tourism Key Lab Digital Innovat Tianfu Culture Chengdu Peoples R China

In multi-criteria decision-making and model evaluation, determining the weight of criteria is crucial. With the rapid development of information technology and the advent of the big data era, the need for complex problem analysis and decision-making has intensified. Traditional CRiteria Importance Through Intercriteria Correlation (CRITIC) methods rely on Pearson correlation, which may not adequately address nonlinearity in some scenarios. This study aims to refine the CRITIC method to better accommodate nonlinear relationships and enhance its robustness. We have developed a novel method named CRiteria Importance Through Intercriteria Dependence (CRITID), which leverages cutting-edge independence testing methods such as distance correlation among others. This approach enhances the assessment of intercriteria relationships. Upon application across diverse data distributions, the CRITID method has demonstrated enhanced rationality and robustness relative to the traditional CRITIC method. These improvements significantly benefit multi-criteria decision-making and model evaluation, providing a more accurate and dependable framework for analyzing complex datasets.

关键词： CRITID CRITIC Independence-zero equivalence property Independence test

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：