检索结果-内蒙古大学图书馆

38th AAAI Conference on Artificial Intelligence, AAAI 2024

作者： Zhu, Yuqi Li, Jia Li, Ge Zhao, YunFei Jin, Zhi Mei, Hong Academy of Military Sciences China Ministry of Education School of Computer Science Peking University Beijing China Advanced Institute of Big Data Beijing China

ISBN: (纸本)1577358872

Recently, Large Language Models (LLMs) have shown impressive abilities in code generation. However, existing LLMs' decoding strategies are designed for Natural Language (NL) generation, overlooking the differences between NL and programming languages (PL). Due to this oversight, a better decoding strategy for code generation remains an open question. In this paper, we conduct the first systematic study to explore a decoding strategy specialized in code generation. With an analysis of loss distributions of code tokens, we find that code tokens can be divided into two categories: challenging tokens that are difficult to predict and confident tokens that can be easily inferred. Among them, the challenging tokens mainly appear at the beginning of a code block. Inspired by the above findings, we propose a simple yet effective method: Adaptive Temperature (AdapT) sampling, which dynamically adjusts the temperature coefficient when decoding different tokens. We apply a larger temperature when sampling for challenging tokens, allowing LLMs to explore diverse choices. We employ a smaller temperature for confident tokens avoiding the influence of tail randomness noises. We apply AdapT sampling to LLMs with different sizes and conduct evaluations on two popular datasets. Results show that AdapT sampling significantly outperforms state-of-the-art decoding strategy. Copyright © 2024, Association for the Advancement of Artificial Intelligence (***). All rights reserved.

关键词： Temperature

来源：评论

学校读者我要写书评

暂无评论

Segment Anything in Medical Images with nnUNet 1

引用

International Challenge on Segment Anything in Medical Images on Laptop held in conjunction with the IEEE/CVF Conference on computer Vision and Pattern Recognition, CVPR 2024

作者： Stock, Raphael Kirchhoff, Yannick Rokuss, Maximilian R. Ravindran, Ashis Maier-Hein, Klaus Heidelberg Germany Faculty of Mathematics and Computer Science Heidelberg University Heidelberg Germany HIDSS4Health - Helmholtz Information and Data Science School for Health Karlsruhe Germany HIDSS4Health - Helmholtz Information and Data Science School for Health Heidelberg Germany Pattern Analysis and Learning Group Department of Radiation Oncology Heidelberg University Hospital Heidelberg Germany

ISBN: (数字)9783031818547

ISBN: (纸本)9783031818530

In this paper, we present an enhanced medical image segmentation approach leveraging the nnUNet framework, specifically tailored to integrate bounding box prompts for improved segmentation accuracy in resource-constrained environments. By incorporating these prompts as binary masks in an additional input channel, we enable more precise and context-aware segmentation. Our methodology employs a 2D slice-wise approach optimized for CPU-based inference through just-in-time (JIT) compiled functions, ensuring efficient processing on standard clinical equipment. Our solution demonstrates robust performance, achieving an average Dice Similarity Coefficient (DSC) of 80.98% and a Normalized Surface Dice (NSD) of 83.23% across multiple modalities in the validation set. This indicates its practical applicability and effectiveness in real-world clinical settings, where computational resources may be limited. By focusing on both accuracy and efficiency, our approach makes advanced segmentation technology accessible to a broader range of healthcare providers, facilitating enhanced clinical decision-making and patient care. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Image segmentation

来源：评论

学校读者我要写书评

暂无评论

Towards robust multimodal sentiment analysis with incomplete data 24

Towards robust multimodal sentiment analysis with incomplete...

引用

Proceedings of the 38th International Conference on Neural Information Processing Systems

作者： Haoyu Zhang Wenbin Wang Tianshu Yu School of Data Science The Chinese University of Hong Kong Shenzhen and Department of Computer Science University College London School of Computer Science Wuhan University School of Data Science The Chinese University of Hong Kong Shenzhen

ISBN: (纸本)9798331314385

The field of Multimodal Sentiment Analysis (MSA) has recently witnessed an emerging direction seeking to tackle the issue of data incompleteness. Recognizing that the language modality typically contains dense sentiment information, we consider it as the dominant modality and present an innovative Language-dominated Noise-resistant Learning Network (LNLN) to achieve robust MSA. The proposed LNLN features a dominant modality correction (DMC) module and dominant modality based multimodal learning (DMML) module, which enhances the model's robustness across various noise scenarios by ensuring the quality of dominant modality representations. Aside from the methodical design, we perform comprehensive experiments under random data missing scenarios, utilizing diverse and meaningful settings on several popular datasets (e.g., MOSI, MOSEI, and SIMS), providing additional uniformity, transparency, and fairness compared to existing evaluations in the literature. Empirically, LNLN consistently outperforms existing baselines, demonstrating superior performance across these challenging and extensive evaluation metrics. The code is available at: https://***/Haoyu-ha/LNLN

关键词：

来源：评论

学校读者我要写书评

暂无评论

Digital Twin-based Dynamic Resource Provisioning Using Deep Q-Network on 6G-enabled Mobile Edge Networks 17

Digital Twin-based Dynamic Resource Provisioning Using Deep ...

引用

17th International Conference on COMmunication Systems and NETworkS, COMSNETS 2025

作者： Alex, Suja A. Singh, Neha Adhikari, Mainak St. Xavier's Catholic College of Engineering Department of Information Technology Nagercoil India Indian Institute of Information Technology Lucknow Department of Computer Science Lucknow India Indian Institute of Science Education and Research Thiruvananthapuram School of Data Science Thiruvananthapuram India

ISBN: (纸本)9798331531195

The advancement of Digital Twin (DT) technology has enabled the creation of digital replicas of physical entitics, significantly enhancing the functionality and performance of mobile networks. However, this innovative approach also introduces considerable overload and performance challenges for mobile edge networks. To address these issucs, we propose a novel DT-enabled edge architecture for 6G-capable technology, employing a dynamic resource allocation strategy. The proposed approach, termed DT-based Dynamic Resource Provisioning (DT-DRP), leverages the capabilities of DT technology and Decp Reinforcement Learning (DRL) to dynamically allocate incoming tasks from mobile devices to edge devices. An intelligent DT server facilitates this allocation, optimizing resource use and improving network performance. The predictive capabilities of the framework are validated using the K-Nearest Neighbor (KNN) dlassifier at the edge network. Experimental results demonstrate that the proposed DT-DRP framework incorporates mobile devices into the DT network scamlessly, achieving superior performance in terms of minimum latency and high accuracy compared to bascline Q-learning algorithms across two medical datasets and various exccution scenarios. This research underscores the potential of DT technology in 6G networks and highlights its efficacy in enhancing mobile edge computing performance through intelligent resource management. © 2025 IEEE.

关键词： Mobile edge computing

来源：评论

学校读者我要写书评

暂无评论

Diverse Geographical Regions Based Biodiversity Conservation by LiDAR Image with Deep Learning Model

引用

Remote Sensing in Earth Systems sciences 2024年第4期7卷 738-749页

作者： Babu C. N, Gireesh Reddy, Chilukala Mahender Kumar, Mungara Kiran Nagaraju, Komatigunta Maheswari, G. Uma Kiran, Ajmeera Department of Computer Science and Engineering BMS Institute of Technology & amp Management Karnataka Bengaluru India Department of Artificial Intelligence and Data Science GITAM (Deemed to be University) Hyderabad India Department of CSE School of Technology GITAM (Deemed to Be University) Hyderabad India Department of Computer Science and Engineering Koneru Lakshmaiah Education Foundation Vaddeswaram Guntur Dist A.P India Department of Computer Science and Engineering RMK College of Engineering and Technology Pudhuvoyal Chennai India Department of Computer Science and Engineering MLR Institute of Technology Hyderabad India

Reliable data on the composition and structure of forests at various spatial scales is necessary for the conservation and monitoring of forest biodiversity. However, because field sampling techniques can be challenging, comprehensive data regarding the features of forest habitat over wide areas are frequently lacking. In order to get over this restriction, we created variables that describe the structure of the forest landscape over a wide environmental gradient using a nationally accessible light detection and ranging (LiDAR) remote sensing dataset. This research proposes a novel technique in forest management based on biodiversity analysis using LiDAR image processing by deep learning techniques. Here, the input LiDAR image has been collected as temperature pattern dataset in major deforestation regions and processed for noise removal and normalization. This image feature has been extracted using recurrent graph learning-based SegNet component analysis. Then, this extracted feature has been classified using Gaussian deep belief region-based adversarial neural networks. The experimental analysis has been carried out for various temperature pattern datasets in terms of random accuracy, average precision, recall, NSE, and AUC. The results of this study will help determine policy objectives for the restoration of mountain areas and enhance the effectiveness of monitoring damaged mountain forests. With the help of the suggested method, 96% random accuracy, 94% average precision, 49% NSE, 93% recall, and 95% AUC were achieved. © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2024.

关键词： Adversarial neural networks Deep learning techniques Deforestation rate LiDAR image processing Temperature rise analysis

来源：评论

学校读者我要写书评

暂无评论

Evaluating Character Understanding of Large Language Models via Character Profiling from Fictional Works

arXiv

引用

arXiv 2024年

作者： Yuan, Xinfeng Yuan, Siyu Cui, Yuhan Lin, Tianhe Wang, Xintao Xu, Rui Chen, Jiangjie Yang, Deqing School of Data Science Fudan University China Shanghai Key Laboratory of Data Science School of Computer Science Fudan University China

Large language models (LLMs) have demonstrated impressive performance and spurred numerous AI applications, in which role-playing agents (RPAs) are particularly popular, especially for fictional characters. The prerequisite for these RPAs lies in the capability of LLMs to understand characters from fictional works. Previous efforts have evaluated this capability via basic classification tasks or characteristic imitation, failing to capture the nuanced character understanding with LLMs. In this paper, we propose evaluating LLMs' character understanding capability via the character profiling task, i.e., summarizing character profiles from corresponding materials, a widely adopted yet understudied practice for RPA development. Specifically, we construct the CROSS dataset from literature experts and assess the generated profiles by comparing them with ground truth references and evaluating their applicability in downstream tasks. Our experiments, which cover various summarization methods and LLMs, have yielded promising results. These results strongly validate the character understanding capability of LLMs. Resources are available at https://***/Joanna0123/character_profiling. Copyright © 2024, The Authors. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Advances, challenges and opportunities in creating data for trustworthy AI (vol 4, pg 669, 2022)

引用

NATURE MACHINE INTELLIGENCE 2022年第10期4卷 904-904页

作者： Liang, Weixin Tadesse, Girmaw Abebe Ho, Daniel Fei-Fei, L. Zaharia, Matei Zhang, Ce Zou, James Department of Computer Science Stanford University Stanford CA USA Department of Biomedical Data Science Stanford University Stanford CA USA IBM Research – Africa Nairobi Kenya Stanford Law School Stanford University Stanford CA USA Department of Computer Science ETH Zurich Zurich Switzerland

As artificial intelligence (AI) transitions from research to deployment, creating the appropriate datasets and data pipelines to develop and evaluate AI models is increasingly the biggest challenge. Automated AI model builders that are publicly available can now achieve top performance in many applications. In contrast, the design and sculpting of the data used to develop AI often rely on bespoke manual work, and they critically affect the trustworthiness of the model. This Perspective discusses key considerations for each stage of the data-for-AI pipeline—starting from data design to data sculpting (for example, cleaning, valuation and annotation) and data evaluation—to make AI more reliable. We highlight technical advances that help to make the data-for-AI pipeline more scalable and rigorous. Furthermore, we discuss how recent data regulations and policies can impact AI.

关键词： Pipelines

来源：评论

学校读者我要写书评

暂无评论

Light Up the Shadows: Enhance Long-Tailed Entity Grounding with Concept-Guided Vision-Language Models

arXiv

引用

arXiv 2024年

作者： Zhang, Yikai He, Qianyu Wang, Xintao Yuan, Siyu Liang, Jiaqing Xiao, Yanghua Shanghai Key Laboratory of Data Science School of Computer Science Fudan University China School of Data Science Fudan University China

Multi-Modal Knowledge Graphs (MMKGs) have proven valuable for various downstream tasks. However, scaling them up is challenging because building large-scale MMKGs often introduces mismatched images (i.e., noise). Most entities in KGs belong to the long tail, meaning there are few images of them available online. This scarcity makes it difficult to determine whether a found image matches the entity. To address this, we draw on the Triangle of Reference Theory and suggest enhancing vision-language models with concept guidance. Specifically, we introduce COG, a two-stage framework with COncept-Guided vision-language models. The framework comprises a CONCEPT INTEGRATION module, which effectively identifies image-text pairs of long-tailed entities, and an EVIDENCE FUSION module, which offers explainability and enables human verification. To demonstrate the effectiveness of COG, we create a dataset of 25k image-text pairs of long-tailed entities. Our comprehensive experiments show that COG not only improves the accuracy of recognizing long-tailed image-text pairs compared to baselines but also offers flexibility and explainability. Copyright © 2024, The Authors. All rights reserved.

关键词： Character recognition

来源：评论

学校读者我要写书评

暂无评论

A federated learning scheme meets dynamic differential privacy

引用

CAAI Transactions on Intelligence Technology 2023年第3期8卷 1087-1100页

作者： Shengnan Guo Xibin Wang Shigong Long Hai Liu Liu Hai Toong Hai Sam State Key Laboratory of Public Big Data College of Computer Science and TechnologyGuizhou UniversityGuiyangChina School of Big Data Key Laboratory of Electric Power Big Data of Guizhou ProvinceGuizhou Institute of TechnologyGuiyangChina School of Information Guizhou University of Finance and EconomicsGuiyangChina Faculty of Business and Communication INTI International UniversityNilaiMalaysia

Federated learning is a widely used distributed learning approach in recent years,however,despite model training from collecting data become to gathering parameters,privacy violations may occur when publishing and sharing models.A dynamic approach is pro-posed to add Gaussian noise more effectively and apply differential privacy to federal deep ***,it is abandoning the traditional way of equally distributing the privacy budget e and adjusting the privacy budget to accommodate gradient descent federation learning dynamically,where the parameters depend on computation derived to avoid the impact on the algorithm that hyperparameters are created *** also incorporates adaptive threshold cropping to control the sensitivity,and finally,moments accountant is used to counting the∈consumed on the privacy‐preserving,and learning is stopped only if the∈_(total)by clients setting is reached,this allows the privacy budget to be adequately explored for model *** experimental results on real datasets show that the method training has almost the same effect as the model learning of non‐privacy,which is significantly better than the differential privacy method used by TensorFlow.

关键词： data privacy machine learning security of data

来源：评论

学校读者我要写书评

暂无评论

Prognostic Phenotype Clustering of Uterine Sarcoma Patients

Prognostic Phenotype Clustering of Uterine Sarcoma Patients

引用

2024 IEEE International Conference on E-Health Networking, Application and Services, HealthCom 2024

作者： Zhang, Tuao Yuan, Suzhen Zhou, Xue Wang, Wenwen Zhu, Xin Graduate School of Computer Science and Engineering The University of Aizu Aizuwakamatsu Japan Department of Obstetrics and Gynecology Tongji Hospital Huazhong University of Science and Technology Wuhan China Department of AI Technology Development M&D Data Science Center Institute of Science Tokyo Tokyo Japan

ISBN: (纸本)9798350350548

Uterine sarcoma is a rare but highly aggressive malignancy with poor prognosis. In this study, prognostic phenotypes corresponding to the end point of mortality were explored by unsupervised machine learning (ML) on training (n = 1985) and validation (n = 498) datasets. The data were extracted from patients with uterine sarcoma in the Surveillance, Epidemiology, and End Results (SEER) database. A decision-making process of phenotype identification was performed and the findings were compared with existing research. Three phenotypes associated with stratified levels of mortality risks (high, intermediate and low) were identified. Kaplan-Meier survival analysis revealed significant differences among these phenotypes (pairwise comparison p © 2024 IEEE.

关键词： Contrastive 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：