检索结果-内蒙古大学图书馆

arXiv 2025年

作者： Cosma, Adrian Cǎtrunǎ, Andy Rǎdoi, Emilian University Politehnica of Bucharest Faculty of Automatic Control and Computer Science Romania

Gait recognition from video streams is a challenging problem in computer vision biometrics due to the subtle differences between gaits and numerous confounding factors. Recent advancements in self-supervised pretraining have led to the development of robust gait recognition models that are invariant to walking covariates. While neural scaling laws have transformed model development in other domains by linking performance to data, model size, and compute, their applicability to gait remains unexplored. In this work, we conduct the first empirical study scaling on skeleton-based self-supervised gait recognition to quantify the effect of data quantity, model size and compute on downstream gait recognition performance. We pretrain multiple variants of GaitPT — a transformer-based architecture — on a dataset of 2.7 million walking sequences collected in the wild. We evaluate zero-shot performance across four benchmark datasets to derive scaling laws for data, model size, and compute. Our findings demonstrate predictable power-law improvements in performance with increased scale and confirm that data and compute scaling significantly influence downstream accuracy. We further isolate architectural contributions by comparing GaitPT with GaitFormer under controlled compute budgets. These results provide practical insights into resource allocation and performance estimation for real-world gait recognition systems. © 2025, CC BY-NC-ND.

关键词： Budget control

来源：评论

学校读者我要写书评

暂无评论

How capable are state-of-the-art language models to cope with sarcasm?

How capable are state-of-the-art language models to cope wit...

引用

International Conference on control Systems and computer science (CSCS)

作者： Alexandru -Costin Băroiu Ştefan Trăuşan-Matu Faculty of Automatic Control and Computer Science Politehnica University of Bucharest Bucharest Romania Romanian Academy Research Institute for Artificial Intelligence “Mihai Drăganescu” Academy of Romanian Scientists Bucharest Romania

Sarcasm detection has established itself as one of the more difficult Natural Language Processing tasks, due to the complex nature of sarcasm. This paper aims to benchmark the performance of state-of-the-art models like BERT, RoBERTa, ALBERT and GPT-3 when faced with this task. The dataset selected is MUStARD, which has increased in popularity in recent years, especially for multimodal tasks, and is one of the most qualitative and data rich dataset. An untuned GPT-3 based model was selected as the baseline and all the other models were fine-tuned using the textual data present in MUStARD, mainly the context and utterance information. The best performer was found to be the GPT-3 fine-tuned model, with an F1 score of 77. This is in line with the reported feats of GPT-3 based models that have popularized in recent months and reaffirms the superiority of GPT-3. Future avenues of research are then presented and explored, and the conclusions are drawn.

关键词：

来源：评论

学校读者我要写书评

暂无评论

automatic detection of vehicles in outdoor parking lots from zenith perspective using neural networks

Automatic detection of vehicles in outdoor parking lots from...

引用

2020 Summer computer Simulation Conference, SCSC 2020, Held at the 2020 Summer Simulation Multi-Conference, SummerSim 2020

作者： García, Juan Manuel Carrera Piorno, Joaquín Recas Mata-Garcia, María Guijarro Department of Computer Architecture and Automatic Control Faculty of Computer Science Universidad Complutense de Madrid Spain

ISBN: (纸本)9781713819462

Nowadays there are a variety of methods to assist parking users in finding free sites in parking lots. However, there is no automatic system that takes into account the size of the car looking for a space or whether the cars adjacent to the free spaces are correctly parked. This paper presents a new method for detecting and calculating the area of vehicles in images taken from a zenith plane using computer vision and machine learning techniques that will help to create a vehicle-oriented search algorithm dedicated to finding the optimal spaces for vehicles entering an outdoor parking lot based on its characteristics. Results with scaled-down and real vehicles show that this new method can detect the area of the vehicles in an image with an average accuracy of 97.98%. ©2020 Society for Modeling & Simulation International (SCS)

关键词： Vehicles

来源：评论

学校读者我要写书评

暂无评论

Onto Digitalization of Public Services

Onto Digitalization of Public Services

引用

International Symposium on Advanced Topics in Electrical Engineering (ATEE)

作者： Ana - Maria Chiriac Stefan Mocanu Cosmin Popa Faculty of Automatic Control and Computer Science UPB Ingenios.ro

A recent report elaborated by European Commission reveals that Romania ranks one of the last positions in European Union in that concerns digitalization of public services. Solutions implemented in other countries (Finland, Estonia, Latvia and many others) prove this is no longer a theoretical concept but a is real need. In this paper, a functional system implemented in several public administration entities from Romania is presented along with the advantages derived from digitalizing public administration services.

关键词： Electrical engineering Data analysis Pandemics Europe computer architecture Computational efficiency Time factors

来源：评论

学校读者我要写书评

暂无评论

Custom-made Near Infrared Spectroscope as a Tool for Obtaining Information Regarding the Brain Condition

Custom-made Near Infrared Spectroscope as a Tool for Obtaini...

引用

International Conference on Methods and Models in Automation and Robotics (MMAR)

作者： Anna Wieczorek Edward Jacek Gorzelańczyk Mariusz Pelc Saravanakumar Duraisamy Luis A. Leiva Aleksandra Kawala-Sterniuk Faculty of Electrical Engineering Automatic Control and Informatics Opole University of Technology Opole Poland Institute of Philosophy Kazimierz Wielki University Bydgoszcz Poland Department of Computer Science University of Luxembourg Luxembourg Luxembourg

We can obtain valuable information about the human brain using functional Near Infrared Spectroscopy (fNIRS). This paper describes the theoretical basis associated with this neuroimaging method through a custom-made prototype of a single-channel fNIRS device. The optodes were soldered to a milled Printed Circuit Board (PCB) and enclosed in a 3D printed housing. Using this fNIRS device, we performed a preliminary study to measure emotional responses from participants. Our results suggest that fNIRS allows for accurate measurement of emotions evoked by positive and negative images.

关键词：

来源：评论

学校读者我要写书评

暂无评论

H control for General Conformable Polynomial Fuzzy Models

引用

International Journal of Fuzzy Systems 2025年 1-10页

作者： Laajimi, Malek Gassara, Hamdi Rhaima, Mohamed Mchiri, Lassaad Makhlouf, Abdellatif Ben Laboratory of Sciences and Techniques of Automatic control & Computer Engineering National School of Engineering of Sfax Sfax Tunisia Department of Statistics and Operations Research College of Science King Saud University Riyadh Saudi Arabia Panthéon-Assas University Paris II Paris France Department of Mathematics Faculty of Sciences Sfax University Sfax Tunisia

This research focuses on exploring the exponential $$H_\infty $$ stability of general conformable nonlinear system. In order to address the nonlinearity inherent in the system, a polynomial fuzzy (PF) method is employed. Modeling a general conformable nonlinear system within the polynomial framework reduces the number of fuzzy rules compared to the classical Takagi–Sugeno fuzzy (TSF). Furthermore, controlling such a complex system, which accounts for perturbations, employs a PF model instead TSF model to describe its nonlinear dynamics, and incorporates a general conformable derivative instead of an integer-order one, is significantly more challenging, and remains unaddressed in previous studies. In this paper, a PF controller is designed in the form of sum of squares (SOS) to enhance the resilience against perturbations and ensure the exponential stability of the proposed model. The proposed SOS can be solved numerically, and partially symbolically, using the recently developed SOSTOOLS. In order to ensure the $$H_\infty $$ performance, a generalized criterion is defined for the general conformable nonlinear system. To demonstrate the effectiveness of the proposed method, a numerical example is provided.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Classification of Skin Lesions from Dermatoscopic Images Using Convolutional Neural Networks

Classification of Skin Lesions from Dermatoscopic Images Usi...

引用

International Conference on control Systems and computer science (CSCS)

作者： Maria Oniga Alina-Elena Sultana Dan Popescu Daniel-Marian Merezeanu Loretta Ichim Faculty of Automatic Control and Computer Science University Politehnica of Bucharest Bucharest Romania Faculty of Electronics Telecommunications and Information Technology University Politehnica of Bucharest Bucharest Romania “St. S. Nicolau” Institute of Virology Bucharest Romania

Skin cancer is one of the most common types of cancer, and it is caused by a variety of dermatological conditions. Identifying abnormalities from skin images is an important pre-diagnostic step to assist physicians in determining the patient’s condition. Thus, to aid dermatologists in the diagnosis process, we proposed five CNN-based classification approaches namely ResNet-101, DenseNet-121, GoogLeNet, VGG16, and MobileNetV2 architectures on which the transfer learning process was applied. The HAM10000-N database consisting of 7,120 images, which was obtained from the original HAM10000 dataset through an augmentation process, was used to train the proposed methods. Moreover, the images from the HAM10000-N were pre-processed by removing hair with the DullRazor algorithm. To evaluate and compare the performance of all networks five metrics were calculated: accuracy, precision, recall, and Fl-score. The best results for the seven-class classification of the HAM10000-N dataset were obtained for DenseNet-121 architecture with 87% accuracy, 0.871 precision, 0.87 recall and 0.872 F1-score.

关键词：

来源：评论

学校读者我要写书评

暂无评论

To be or not to be? Assessment on using touchscreen as inceptor in flight operation

引用

Transportation Research Procedia 2022年 66卷 117-124页

作者： Wen-Chin Li Yifan Wang Wojciech Tomasz Korek Safety and Accident Investigation Centre Cranfield University Cranfield United Kingdom Dynamics Simulation and Control Group Cranfield University Cranfield United Kingdom Faculty of Automatic Control Electronics and Computer Science Silesian University of Technology Poland

Introduction . The innovative concept on applying touchscreen controls on the flight deck design had been discussed for a long time. However, there are some potential risks on touchscreen applications constrained by the issues associated with turbulence and pilots’ inadvertent activation. Research questions . This research aims to evaluate human-computer interactions and handling quality using touchscreens as inceptor in flight operations. Method . The scenario was set to conduct an instrument landing on the final approach using Future System Simulator (FSS). There are 8 commercial pilots (flight hours M = 4475.0, SD = 2742.1) using three different inceptors including traditional sidestick, touchscreen and gamepad for ILS landing. Results . There was a significant difference among three inceptors on handling quality in both landing without turbulence (F (2,14) = 6.25, p =.01, η p 2  = .47) and landing with turbulence (F (2,14) = 3.93, p =.04, η p 2  = .36) scenarios. Furthermore, post Hoc comparisons revealed that the handling quality of touchscreen was significantly lower than sidestick and gamepad. Discussion . By analyzing participants’ empirical experiences, the touchscreen inceptor was rated as the lowest handling quality among three inceptors due to the novel and lack of practice effects in flight operations. However, there is a potential on the information supply for touchscreen inceptor based on pilots’ feedbacks. Conclusion . Touchscreens provide numerous benefits for making flight decks simpler, but the usage as an inceptor is still in its infancy and there are still lots of problems that need to be fixed. Future Systems Simulator (FSS) is a highly reconfigurable modular flight simulator that allows pilots/researchers to explore the potential on future flight decks design for single pilot operations. There are some potential benefits on the implementation touchscreen inceptor for future flight deck design if the human-centred design principle can be integr

关键词： Flight Deck Design Human-computer Interactions Handling Quality Turbulence Situation Awareness

来源：评论

学校读者我要写书评

暂无评论

Handwritten Signature Recognition using Parallel CNNs and Transfer Learning for Forensics

Handwritten Signature Recognition using Parallel CNNs and Tr...

引用

International Conference on control, Decision and Information Technologies (CoDIT)

作者： Jaouhar Fattahi Feriel Sghaier Mohamed Mejri Ridha Ghayoula Emil Pricop Baha Eddine Lakdher Department of Computer Science and Software Engineering Laval University Quebec Canada Carthage National Engineering School Carthage University Tunis Tunisia Faculty of Engineering University of Moncton Canada Computers & Electronics Department Automatic Control Petroleum-Gas University of Ploieşti Ploieşti Romania Faculty of Science University of Monastir Monastir Tunisia

ISBN: (数字)9798350373974

ISBN: (纸本)9798350373981

Handwritten signatures hold paramount importance in legal, financial, and administrative domains, necessitating the development of robust signature recognition tools for forensic applications. This paper introduces a handwritten signature recognition (HSR) model employing Parallel Convolutional Neural Networks (CNN) tailored for forensic endeavors. Utilizing the parallel processing capabilities of CNN, our proposed approach adeptly analyzes and extracts discriminative features from handwritten signature images to facilitate precise recognition. In addition, we leverage several transfer learning techniques by parallelizing proven pre-trained CNNs. Extensive experimentation validates the efficacy of our approach on a standard dataset, demonstrating high accuracy and resilience in signature recognition tasks. The proposed approach exhibits substantial promise in augmenting forensic investigations by automating signature verification processes, thereby bolstering fraud detection efforts and upholding the integrity of legal documentation.

关键词： Deep learning Handwriting recognition Forensics Face recognition Transfer learning Feature extraction Fraud Convolutional neural networks Standards Resilience

来源：评论

学校读者我要写书评

暂无评论

BigFeat: Scalable and Interpretable Automated Feature Engineering Framework

BigFeat: Scalable and Interpretable Automated Feature Engine...

引用

IEEE International Conference on Big Data

作者： Hassan Eldeeb Shota Amashukeli Radwa ElShawi Institute of Computer Science University of Tartu Tartu Estonia Computers and Automatic Control Department Faculty of Engineering Tanta University Tanta Egypt.

ISBN: (纸本)9781665480468

Feature engineering is a crucial step in building well-performing machine learning pipelines. However, manually constructing highly predictive features is time-consuming and requires domain knowledge. Although the research area of automated feature engineering has attracted much interest lately, both in academia and industry, the scalability and efficiency of the existing systems and tools are still practically unsatisfactory. This paper presents a scalable and interpretable automated feature engineering framework, BigFeat, that optimizes input features’ quality to maximize the predictive performance according to a user-defined metric. BigFeat employs a dynamic feature generation and selection mechanism that constructs a set of expressive features that improve the prediction performance while retaining interpretability. Extensive experiments are conducted, and the results show that BigFeat provides superior performance compared to the state-of-the-art automated feature engineering framework, AutoFeat, on a wide range of datasets. We show that BigFeat significantly improves the F1-Score of 8 classifiers by 4.59%, on average. In addition, the performance improvement achieved by integrating BigFeat into different AutoML frameworks is higher than that achieved by integrating AutoFeat into the same frameworks. Besides, the scalability of BigFeat is confirmed by its linear complexity, parallel design, and execution time which is, on average, 22x faster than AutoFeat.

关键词： Measurement Industries Scalability Pipelines Buildings Machine learning Big Data

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：