检索结果-内蒙古大学图书馆

arXiv 2025年

作者： Chen, Zhe Chen, Zijing Cisco-La Trobe Centre for Artificial Intelligence and Internet of Things Department of Computer Science and Information Technology La Trobe University Edwards Rd VIC3552 Australia

Object Referring Analysis (ORA), commonly known as referring expression comprehension, requires the identification and localization of specific objects within an image based on natural language descriptions. Unlike conventional object detection, ORA requires both accurate language understanding and precise visual localization, making it inherently more complex. Although recent pre-trained large visual grounding detectors have achieved significant progress, these approaches heavily rely on extensively labeled data for fine-tuning and can be burdened by time-consuming learning processes. To address these limitations, we introduce a novel, training-free framework for zero-shot ORA, termed FLORA (Formal language for Object Referring and Analysis). FLORA harnesses the inherent reasoning capabilities of large language models (LLMs) and integrates a formal language model - a logical framework that regulates natural language within structured, rule-based descriptions - to provide effective zero-shot ORA. In particular, with the help of LLMs, our formal language model (FLM) enables an effective, logic-driven interpretation of object descriptions without necessitating any training processes. Built upon FLM-regulated LLM outputs, we further devise a Bayesian inference framework and employ appropriate off-the-shelf interpretive models to finalize the reasoning, delivering favorable robustness against LLM hallucinations and compelling ORA performance in a training-free manner. In practice, our FLORA offers significant improvements over existing pretrained grounding detectors, boosting their zero-shot performance by up to around 45%. Our comprehensive evaluation across different challenging datasets also confirms that FLORA consistently surpasses current state-of-the-art zero-shot methods, providing superior performance in both detection and segmentation tasks associated with zero-shot ORA. We believe our probabilistic parsing and reasoning of the LLM outputs elevate the reliabil

关键词： Modeling languages

来源：评论

学校读者我要写书评

暂无评论

A Deep Learning Knowledge Graph Neural Network for Recommender Systems

SSRN

引用

SSRN 2023年

作者： Kaur, Gurinder Liu, Fei Chen, Yi-Ping Phoebe Department of Computer Science and Information Technology La Trobe University VIC Australia

Knowledge graphs are becoming the new state-of-the-art for recommender systems. This paper is based on knowledge graphs to alleviate the problem of data sparsity. Various methods have been recently deployed to solve this problem which largely attempts to study user-item representation and then recommend items to users based on these representations. Although these methods are effective, they lack explainability for recommendations and do not mine side information. In this paper, we propose the use of knowledge graphs which includes additional information about users and items in addition to the use of a user/item interaction matrix. The vital element of our model is neighbourhood aggregation for collaborative filtering. Every user and item are associated with an ID embedding, which is circulated on the interaction graph for users, items, and their attributes. We obtain the final embeddings by combining the embeddings learned at various hidden layers with a biased sum. Our model is easier to train and achieves better performance compared to graph neural network-based collaborative filtering (GCF) and other state-of-the-art recommender methods. We provide evidence for our argument by analytically comparing the knowledge graph convolution network (KGCN) with GCF and eight other state-of-the-art methods, using similar experimental settings and the same datasets. © 2023, The Authors. All rights reserved.

关键词： Recommender systems

来源：评论

学校读者我要写书评

暂无评论

Multimodal Supported Chatbot Framework for People with Aphasia

Multimodal Supported Chatbot Framework for People with Aphas...

引用

Moratuwa engineering Research Conference (MERCon)

作者： R.L.A. Sanduni Ayesha L.B. Hasini M. Vijayarathna N.P.A. Vithanage Uthayasanker Thayasivam Damminda Alahakoon Achini Adikari Nuwan Pallewela Department of Computer Science and Engineering University of Moratuwa Katubedda Sri Lanka Research Centre for Data Analytics and Cognition La Trobe University Australia

ISBN: (数字)9798331529048

ISBN: (纸本)9798331529055

Chatbots have become a trending topic with emerging platforms like ChatGPT, Gemini, and Copilot, for conversation assistance. Current chatbots mainly focus on the general public assuming a natural flow of conversation. However, there is a need for a chatbot that supports people with various communication disabilities. This research fills this gap by offering a novel technique for a chatbot that assists people with Aphasia, a condition characterised by difficulties with language. We propose a multi-modal chatbot that is customised and designed to assist users with communication disabilities in navigating a website. Unlike typical chatbots, which rely on one form of communication, our architecture combines multiple modalities to improve comprehension and promote effective communication for people with Aphasia. We focus on gathering multimodal inputs by recognising and combining user intents from diverse sources. The use of Txtai, an all-in-one embeddings database for semantic search improves our chatbot’s capacity to process various inputs efficiently. We leverage specialised models like Whisper for audio transcription and MediaPipe Gesture Recognizer for gesture detection to enhance user interactions. Additionally, Rasa Core integration improves conversational experiences for users. We propose that this new approach will make communication more accessible and inclusive for individuals with Aphasia.

关键词： Navigation Semantic search Intent recognition Diversity reception Oral communication Gesture recognition Machine learning Aphasia Chatbots User experience

来源：评论

学校读者我要写书评

暂无评论

Emotion-oriented Cross-modal Prompting and Alignment for Human-centric Emotional Video Captioning

引用

IEEE Transactions on Multimedia 2025年 27卷 3766-3780页

作者： Wang, Yu Liu, Yuanyuan Zhou, Shunping Huang, Yuxuan Tang, Chang Zhou, Wujie Chen, Zhe China University of Geosciences School of Computer Science School of Geography and Information Engineering Wuhan430074 China China University of Geosciences School of Computer Science Wuhan430074 China Zhejiang University of Science and Technology School of Information and Electronic Engineering Hangzhou310018 China La Trobe University Cisco-La Trobe Centre for AI and IoT School of Computing Engineering and Mathematical Sciences BundooraVIC3086 Australia

Human-centric Emotional Video Captioning (H-EVC) aims to generate fine-grained, emotion-related sentences for human-based videos, enhancing the understanding of human emotions and facilitating human-computer emotional interaction. However, existing video captioning methods primarily focus on overall event content, often overlooking sufficient subtle emotional clues and interactions in videos. As a result, the generated captions frequently lack emotional information. To address this, we propose a novel Emotion-oriented Cross-modal Prompting and Alignment (ECPA) approach for large foundation models to enhance H-EVC accuracy by effectively modeling fine-grained visual-textual emotion clues and interactions. Using large foundation models, our ECPA introduces two learnable prompting strategies: visual emotion prompting (VEP) and textual emotion prompting (TEP), as well as an emotion-oriented cross-modal alignment (ECA) module. In VEP, we develop two-level learnable visual prompts, i.e., emotion recognition (ER)-level and action unit (AU)-level prompting, to assist pre-trained vision-language foundation models to attend to both coarse and fine emotion-related visual information in videos. In TEP, we correspondingly devise two-level learnable textual prompts, i.e., sentence-level emotional tokens, and word-level masked tokens, for obtaining both whole and local textual prompt representations related to emotions. To further facilitate the interaction and alignment of visual-textual emotion prompt representations, our ECA introduces another two levels of emotion-oriented prompt alignment learning mechanisms: the ER-sentence level and the AU-word level alignment losses. Both enhance the model's ability to capture and integrate both global and local cross-modal emotion semantics, thereby enabling the generation of fine-grained emotional linguistic descriptions in video captioning. Extensive experiments not only demonstrate that our ECPA outperforms existing state-of-the-art ap

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

Developing an Uncertainty-Aware Empathetic Conversational System

Developing an Uncertainty-Aware Empathetic Conversational Sy...

引用

International Conference on Affective Computing and Intelligent Interaction Workshops and Demos (ACIIW)

作者： Samad Roohi Department of Computer Science and Information Technology La Trobe University Melbourne Australia

This document is a model and instructions for laTEX. Natural empathetic dialogue requires an understanding of emotion, intention, and context. In empathetic conversational systems, accurate classification of emotion, intention, and understanding of context contribute to a more natural and effective dialogue between the system and the user. However, due to the subjective nature of these concepts and the lack of contextual cues, the resulting models are susceptible to uncertain outputs, causing misunderstanding and inappropriate response generation in empathetic conversational systems. This research proposal aims to address this issue by proposing the use of Bayesian inference methods to estimate and reduce uncertainty in emotion, intention, and context modeling tasks. By incorporating uncertainty estimates into the downstream task of response generation, the proposed uncertainty-aware approach is expected to improve the performance of empathetic dialogue generation tasks significantly. The potential impact of this research is providing a more natural, effective, and engaging user experience in conversational systems, with applications in areas such as mental health, customer service, and education.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Personalized Federated Learning via ADMM with Moreau Envelope

arXiv

引用

arXiv 2023年

作者： Zhu, Shengkun Zeng, Jinshan Wang, Sheng Sun, Yuan Peng, Zhiyong The School of Computer Science Wuhan University China The School of Computer and Information Engineering Jiangxi Normal University Nanchang China La Trobe Business School La Trobe University Australia

Personalized federated learning (PFL) is an approach proposed to address the issue of poor convergence on heterogeneous data. However, most existing PFL frameworks require strong assumptions for convergence. In this paper, we propose an alternating direction method of multipliers (ADMM) for training PFL models with Moreau envelope (FlaME), which achieves a sublinear convergence rate, relying on the relatively weak assumption of gradient Lipschitz continuity. Moreover, due to the gradient-free nature of ADMM, FlaME alleviates the need for hyperparameter tuning, particularly in avoiding the adjustment of the learning rate when training the global model. In addition, we propose a biased client selection strategy to expedite the convergence of training of PFL models. Our theoretical analysis establishes the global convergence under both unbiased and biased client selection strategies. Our experiments validate that FlaME, when trained on heterogeneous data, outperforms state-of-the-art methods in terms of model performance. Regarding communication efficiency, it exhibits an average speedup of 3.75× compared to the baselines. Furthermore, experimental results validate that the biased client selection strategy speeds up the convergence of both personalized and global models. © 2023, CC BY-NC-SA.

关键词： Machine learning

来源：评论

学校读者我要写书评

暂无评论

On the Prospects of Stereolithography Method for 3D Printing of Future Microwave Devices with Low-Dk

On the Prospects of Stereolithography Method for 3D Printing...

引用

2023 IEEE International Symposium on Antennas and Propagation and USNC-URSI Radio science Meeting, AP-S/URSI 2023

作者： Mitra, Dipankar Striker, Ryan Nusrat, Tasin Gadhamshetty, Venkataramana Braaten, Benjamin D. Roy, Sayan University of Wisconsin-La Crosse Department of Computer Science La CrosseWI United States College of Engineering and Mines University of North Dakota Grand ForksND United States South Dakota Mines Department of Electrical Engineering and Computer Science Rapid CitySD United States South Dakota Mines Department of Civil and Environmental Engineering Rapid CitySD United States North Dakota State University Department of Electrical and Computer Engineering FargoND United States

ISBN: (纸本)9781665442282

The stereolithography (Sla) method is known for its high-quality surface finish and details, hence, gaining its popularity with the realization of 3D-printed parts and structures for microwave applications. On the other hand, low-dielectric (Dk) materials have an utmost importance in high-frequency PCB design and microwave applications. Here, the authors present the prospects of the Sla method in 3D-printing of Low-Dk materials for future microwave devices. The Sla method is a relatively new manufacturing process, so it is critical to explore a range of dielectric constants of a few prospective commercially available materials. Here, eleven different resin-based materials were 3D-printed as cubes using the Sla method and characterized in terms of dielectric constant (ϵr) and loss tangent (tanδ) using broadband coaxial probe method. The measured dielectric constant values were within the range of 2 - 4 for all the tested materials which thus proved to be among promising candidates for realizing microwave devices in the frequency bandwidth of 1 to 20 GHz. A few observed failures while manufacturing certain other shapes were also reported as a limitation of the Sla method. © 2023 IEEE.

关键词： Stereolithography

来源：评论

学校读者我要写书评

暂无评论

Development and Characterization of Zinc Dry Electrodes for Wearable Electrophysiology 46

Development and Characterization of Zinc Dry Electrodes for ...

引用

46th Annual International Conference of the IEEE engineering in Medicine and Biology Society, EMBC 2024

作者： Rizq, Cassia D'Amico, Alessandro Truel, Aidan Faybishenko, Joelle Lee, Min Suk Kim, Jeong-Hoon Cauwenberghs, Gert De Sa, Virginia R. University of California San Diego Department of Bioengineering La Jolla United States University of California San Diego Department of Cognitive Science La Jolla United States University of California San Diego Halicioǧlu Data Science Institute La Jolla United States University of California San Diego Department of Electrical and Computer Engineering La Jolla United States

ISBN: (纸本)9798350371499

Zinc dry electrodes were fabricated and investigated for wearable electrophysiology recording. Results from electrochemical impedance spectroscopy and electromyography functionality testing show that zinc electrodes are suitable for use in electrophysiology. Two electrode configurations were tested: a standard disc and a custom tripolar concentric ring configuration. However, no functional benefit was observed with the tripolar concentric ring electrodes as compared to the disc electrodes. © 2024 IEEE.

关键词： Electromyography

来源：评论

学校读者我要写书评

暂无评论

Integration of Satellite Monitoring data in a Digital Twin of Transport Infrastructure 15

Integration of Satellite Monitoring data in a Digital Twin o...

引用

Earth Resources and Environmental Remote Sensing/GIS Applications XV 2024

作者： Napolitano, Antonio Gagliardi, Valerio D’Amico, Fabrizio Benedetto, Andrea Department of Civil Computer Science and Aeronautical Engineering Roma Tre University Via Vito Volterra 62 Rome00146 Italy Department of Civil Constructional and Environmental Engineering University La Sapienza Via Eudossiana 18 Rome00184 Italy

ISBN: (纸本)9781510681026

In recent years, considerable attention has been directed towards the management of transportation infrastructure by stakeholders and governing bodies, aimed at adopting more effective and efficient technologies for monitoring and maintenance purposes. A comprehensive analysis of the health conditions of transportation assets and their surrounding environment plays a decisive role in identifying the factors influencing these infrastructures, including natural hazards and gradual morphological changes. Understanding environmental conditions, such as subsidence or landslide phenomena near transportation assets, is crucial for accurately predicting infrastructure variations over time and intervening promptly to prevent or efficiently repair damages. However, the monitoring of such phenomena (e.g., landslides) is generally not conducted at the network-scale level to evaluate the effects and potential risks on infrastructure assets. In this context, satellite remote sensing techniques, particularly utilizing multi-source and multi-frequency satellite missions, are increasingly applied for their capability to regularly monitor transport assets and extensive areas near their locations. This research explores the integration of information derived from satellite data, proposing a novel methodology for managing multi-sensor survey data by integrating satellite remote sensing into a dynamic Digital Twin of an infrastructure. To achieve this, Synthetic Aperture Radar (SAR) data were processed using Multi-Temporal SAR Interferometry (MTInSAR) to detect potential damages and changes in transport infrastructures and their surrounding environment. An experimental application focusing on a bridge was developed for this purpose. Data obtained from satellite remote sensing missions, including Sentinel-1 and COSMO-SkyMed missions, were integrated into the Digital Twin. A digital modeling process was developed to merge these datasets, resulting in the creation of a comprehensive Digita

关键词： Network security

来源：评论

学校读者我要写书评

暂无评论

An improved Spectral Analysis based Subspace Detection for Hyperspectral Image Classification

An improved Spectral Analysis based Subspace Detection for H...

引用

International Conference on computer and Information Technology (ICCIT)

作者： Kazi Mehrab Rashid Md. Ali Hossain Tasmia Jannat Md. Anisur Rahman Deptartment of Computer Science & Engineering Rajshahi University of Engineering & Technology Bangladesh Department of Business Analytics La Trobe University Australia

Hyperspectral imagery, with its rich spectral information, is indispensable in various remote sensing applications. However, working with hyperspectral data introduces two formidable challenges: the curse of dimensionality and the predicament of class imbalance. The inherent high dimensionality necessitates substantial computational resources, while class imbalance undermines overall classification accuracy due to the scarcity of samples in training for certain classes. To mitigate the curse of dimensionality, this research analyses a few diverse feature extraction techniques, including Principal Component Analysis (PCA), Sparse PCA, segmented PCA, and Linear Discriminant Analysis (LDA). Although these methods can reduce the features from the original dataset, however sometimes depend on certain global statistics such as variance. Also, these methods could not handle the small classes as they could not represent the overall characteristics of the samples. Therefore, the proposed method incorporates the solution to the class imbalance problem as well as reduces the features based on the given classes. The ensuing classification phase employs a kernel Support Vector Machine (KSVM), achieving an accuracy of 88.81% only when the class imbalance exists. later the class imbalances are resolved through the incorporation of Synthetic Minority Over-sampling Technique (SMOTE). The proposed approach reduces the input features and generates synthetic data for each class based on the proportions original simultaneously rectifying class imbalance while preserving dataset characteristics. Remarkably, this strategy results in an improvement of accuracy to 98.69%. This study underscores the significance of hyperspectral imagery, delineates its inherent challenges, and presents novel solutions that substantially enhance classification performance, making hyperspectral data more accessible and valuable in remote sensing endeavors.

关键词：

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：