检索结果-内蒙古大学图书馆

Joint 30th International Conference on Computational Linguistics and 14th International Conference on Language Resources and Evaluation, LREC-COLING 2024

作者： Yin, Wen Liu, Cencen Xu, Yi Wahla, Ahmad Yiting, Huang Zheng, Dezhang School of Information and Software Engineering University of Electronic Science and Technology of China Chengdu610054 China Laboratory of Intelligent Collaborative Computing Chengdu611731 China Bashu Middle School Chongqing400013 China

ISBN: (纸本)9782493814104

Although there have been some studies uses prompt learning for the Aspect-based Sentiment Analysis(ABSA) tasks, their methods of prompt-tuning are simple and crude. Compared with vanilla fine-tuning methods, prompt learning intuitively bridges the objective form gap between pre-training and fine-tuning. Concretely, simply constructing prompt related to aspect words fails to fully exploit the potential of Pre-trained Language Models, and conducting more robust and professional prompt engineering for downstream tasks is a challenging problem that needs to be solved urgently. Therefore, in this paper, we propose a novel syntax-aware enhanced prompt method (SynPrompt), which sufficiently mines the key syntactic information related to aspect words from the syntactic dependency tree. Additionally, to effectively harness the domain-specific knowledge embedded within PLMs for the ABSA tasks, we constructed two adaptive prompt frameworks to enhance the perception ability of the above method. After conducting extensive experiments on three benchmark datasets, we have found that our method consistently achieves favorable results. These findings not only demonstrate the effectiveness and rationality of our proposed methods but also provide a powerful alternative to traditional prompt-tuning. © 2024 ELRA Language Resource Association: CC BY-NC 4.0.

关键词： Sentiment analysis

来源：评论

学校读者我要写书评

暂无评论

A Lightweight and High-Fidelity Model for Generalized Audio-Driven 3D Talking Face Synthesis 27th

A Lightweight and High-Fidelity Model for Generalized Audi...

引用

27th International Conference on Pattern Recognition, ICPR 2024

作者： Liu, Shunce Zhong, Yuwei Wang, Huixuan Peng, Jingliang Shandong Provincial Key Laboratory of Network Based Intelligent Computing University of Jinan Jinan China School of Information Science and Engineering University of Jinan Jinan China School of Software and Microelectronics Peking University Beijing China School of Software Shandong University Jinan China

ISBN: (纸本)9783031781247

We present LW-GeneFace, a lightweight and high-fidelity model for generalized audio-driven facial animation, in this paper. We develop this model by reducing the size while maintaining the synthetic quality of GeneFace, an audio-driven facial animation model known for its high fidelity and generalization capabilities. Specifically, we compress the first and the third stages of GeneFace as they dominate the model size. In the first stage, we propose a lightweight version of the WaveNet-based network inspired by MobileNetV3 and DP-block. It utilizes depthwise separable convolution and dual-path feature extraction to compress the network while maintaining effective feature extraction. The shared network structure in the dual-path feature extraction further reduces model complexity and improves training efficiency. In the third stage, we generate realistic 3D renderings at reduced model size by introducing novelties in RAD-NeRF. Technically, we reduce the hash table sizes in the grid-based encoding modules, as well as present a lightweight bottleneck MLP architecture to increase the non-linearity of the model. Experimental results demonstrate that LW-GeneFace achieves state-of-the-art performance with both model size and synthetic quality considered. The source code of LW-GeneFace will be released after acceptance of this paper. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Animation

来源：评论

学校读者我要写书评

暂无评论

Robust Imagined Speech Production Using AI-Generated Content Network for Patients with Language Impairments

引用

IEEE Transactions on Consumer Electronics 2024年第1期71卷 1402-1411页

作者： Xu, Xu Fu, Chong Dalian University of Technology School of Software Dalian116621 China Northeastern University School of Computer Science and Engineering Shenyang110004 China Northeastern University Key Laboratory of Intelligent Computing in Medical Image Ministry of Education Shenyang110819 China Ministry of Education Engineering Research Center of Security Technology of Complex Network System Shenyang110004 China

Imagined speech production is critical for brain-computer interface systems. It is able to provide the communication ability for patients with language impairments. Nowadays, many studies have developed algorithms for this purpose. However, they often struggle with an issue that is the noise in physiological signals from various devices. This makes current methods have low robustness when performing speech generation. To this end, we propose a framework to robustly producing imagined speech for patients with language impairments. It is based on artificial intelligent-generated content (AIGC). In specific, this work develops a diffusion model to remove noise from brain signals, which are then converted and normalized into log-mel spectrograms. Next, the product feeds into a transformer-based model to extract local and global features. Finally, these enhanced features are inputted into full connected layers to generate the imagined speech. The effectiveness of our model is validated using the high-quality, i.e., single-word-production-Dutch-iBIDS dataset. It achieves Pearson correlation scores above 0.8 and standard deviations below 0.2 across different volunteers. Experimental results well demonstrate that our model is robust and has advantages over peer methods. © 2024 IEEE.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

Learning a Coordinated Network for Detail-Refinement Multiexposure Image Fusion

引用

IEEE Transactions on Circuits and Systems for Video Technology 2023年第2期33卷 713-727页

作者： Li, Jiawei Liu, Jinyuan Zhou, Shihua Zhang, Qiang Kasabov, Nikola K. Ministry of Education School of Software Engineering Dalian University Key Laboratory of Advanced Design and Intelligent Computing Dalian116622 China Dalian University of Technology School of Software Technology Dalian116024 China Auckland University of Technology Knowledge Engineering and Discovery Research Institute Auckland1010 New Zealand Ulster University Intelligent Systems Research Center LondonderryBT52 1SA United Kingdom

Nowadays, deep learning has made rapid progress in the field of multi-exposure image fusion. However, it is still challenging to extract available features while retaining texture details and color. To address this difficult issue, in this paper, we propose a coordinated learning network for detail-refinement in an end-to-end manner. Firstly, we obtain shallow feature maps from extreme over/under-exposed source images by a collaborative extraction module. Secondly, smooth attention weight maps are generated under the guidance of a self-attention module, which can draw a global connection to correlate patches in different locations. With the cooperation of the two aforementioned used modules, our proposed network can obtain a coarse fused image. Moreover, by assisting with an edge revision module, edge details of fused results are refined and noise is suppressed effectively. We conduct subjective qualitative and objective quantitative comparisons between the proposed method and twelve state-of-the-art methods on two available public datasets, respectively. The results show that our fused images significantly outperform others in visual effects and evaluation metrics. In addition, we also perform ablation experiments to verify the function and effectiveness of each module in our proposed method. The source code can be achieved at https://***/lok-18/LCNDR. © 1991-2012 IEEE.

关键词： Extraction

来源：评论

学校读者我要写书评

暂无评论

A Text Structure-based Extractive And Abstractive Summarization Method 7

A Text Structure-based Extractive And Abstractive Summarizat...

引用

7th International Conference on intelligent computing and Signal Processing, ICSP 2022

作者： Yan, Jing Zhou, Shihua Dalian University Key Laboratory of Advanced Design and Intelligent Computing Ministry of Education School of Software Engineering Dalian China

ISBN: (纸本)9781665478571

Extraction summarization and abstraction summarization have advantages and disadvantages, so how to better combine these two ways has become a difficult problem. To address this challenge, this paper proposed a new fusion method. The major novelty lies in the design of the new method. Wherein our approach, first, use the text structure information and the idea of the K-means algorithm to divide the text into regions, and then the main part and the non-main parts are determined according to the distribution of the subject words. Next, apply information extraction on the main part and text generation on the non-main parts. Finally, the two parts of the summarization are merged according to the sequence of the text. Experimental results show that the quality of the summarization is better than that of extraction summarization and abstraction summarization. In addition, to make the method more targeted in Chinese text processing, the Cw2vec model based on Chinese stroke information is used in the encoding process, and the experiment proves that the quality of summarization can be further improved. © 2022 IEEE.

关键词： Text processing

来源：评论

学校读者我要写书评

暂无评论

Byzantine-Robust Federated Learning on Non-IID Data via Inversing Artificial Gradients

Byzantine-Robust Federated Learning on Non-IID Data via Inve...

引用

2024 IEEE Global Communications Conference, GLOBECOM 2024

作者： Li, Minyang Tang, Xiangyun Zhang, Tao Liu, Guangyuan Weng, Yu Minzu University of China Key Laboratory of Ethnic Language Intelligent Analysis and Security Governance China Beijing Jiaotong University School of Software Engineering China Nanyang Technological University College of Computing and Data Science Singapore

ISBN: (纸本)9798350351255

Federated Learning (FL) is a distributed machine learning framework that enhances privacy by enabling multiple participants to train a global model without sharing their raw data. However, FL still faces the threat posed by data and Byzantine attacks (e.g., data poisoning and model poisoning attacks) from malicious clients aimed to decrease the FL global training accuracy. Previous studies have proposed solutions to improve the robustness of FL systems. However, these robust approaches have not effectively defended against Byzantine attacks from malicious clients in non-Independent and Identically Distributed (non-IID) environments, decreasing overall training accuracy. In this work to address this issues, we propose a new defence framework that aims to enhance model accuracy against Byzantine attacks from malicious clients in non-IID environments. In our approach, the central server performs the Inverse Deep Learning Gradient attack using the gradient data submitted by users in each round, obtaining an inversed artificial gradient. We then assess the squared L2 norm difference between this inversed gradient and the actual gradients to detect malicious clients. Simultaneously, we assign weights to each client involved in the aggregation based on differences in the squared L2 norm, aiming to minimize the impact of malicious actors on the overall model. We have experimentally evaluated our method and demonstrated its ability to maintain training accuracy under attack in non-IID environments using standard datasets. © 2024 IEEE.

关键词： Federated learning

来源：评论

学校读者我要写书评

暂无评论

Review of Inference Time Prediction Approaches of DNN: Emphasis on Service robots with cloud-edge-device architecture

Review of Inference Time Prediction Approaches of DNN: Empha...

引用

2023 IEEE International Conference on Robotics and Biomimetics, ROBIO 2023

作者： Xiang, Tian Meng, Qiwei Zhang, Ji Zhang, Beibei Song, Wei Xie, Anhuan Gu, Jason Research Institute of Interdisciplinary Innovation Zhejiang Laboratory Research Center for Intelligent Robotics Hangzhou311100 China The Chinese University of Hong Kong T Stone Robotics Institute Department of Mechanical and Automation Engineering Hong Kong Research Institute of Intelligent Computing Zhejiang Laboratory Research Center for Intelligent Computing Software Hangzhou311100 China Dalhousie University Electrical and Computer Engineering Halifax Canada

ISBN: (纸本)9798350325706

In recent years, the global robot market has witnessed substantial growth, particularly in the domain of service robots. Despite their expanding presence, service robots encounter limitations when operating autonomously in unstructured environments, primarily due to their constrained computational capacities. As a solution, the fusion of cloud and edge computing resources becomes imperative to expedite task inference and enhance scenario perception capabilities. The integration of cloud-edge-device models holds significant promise in bolstering the operational efficiency of robots. This entails the dynamic partitioning of intricate robotic tasks, executed collaboratively across cloud, edge, and device resources. In this landscape, deep neural network (DNN) models play a pivotal role in facilitating a wide array of robotic tasks. The inference time for each layer of a DNN model in actual deployment, emerges as a critical determinant in the model's partitioning strategy. It also serves as an important metric influencing the model's suitability for a specific hardware platform. This article presents an overview of recent advancements in predicting inference and training time of DNN models, summarizes the related methods, and finally discusses the challenges in this field and the research that can be studied in the future. © 2023 IEEE.

关键词：

来源：评论

学校读者我要写书评

暂无评论

GAN-based Symmetric Embedding Costs Adjustment for Enhancing Image Steganographic Security 24

GAN-based Symmetric Embedding Costs Adjustment for Enhancing...

引用

32nd ACM International Conference on Multimedia, MM 2024

作者： Ye, Miaoxin Zhou, Saixing Luo, Weiqi Tan, Shunquan Huang, Jiwu Guangdong Provincial Key Laboratory of Information Security Technology Sun Yat-Sen University Guangzhou China Shenzhen Key Laboratory of Media Security College of Computer Science and Software Engineering Shenzhen University Shenzhen China Guangdong Laboratory of Machine Perception and Intelligent Computing Faculty of Engineering Shenzhen MSU-BIT University Shenzhen China

ISBN: (纸本)9798400706868

Designing embedding costs is pivotal in modern image steganography. Many studies have shown adjusting symmetric embedding costs to asymmetric ones can enhance steganographic security. However, most existing methods heavily depend on manually defined parameters or rules, limiting security performance improvements. To overcome this limitation, we introduce an advanced GAN-based framework that transitions symmetric costs to asymmetric ones without the need for the manual intervention seen in existing approaches, such as the detailed specification of cost modulation directions and magnitudes. In our framework, we firstly achieve symmetric costs for a cover image, which is randomly split into two sub-images, with part of the secret information embedded into one. Subsequently, we design a GAN model to adjust the embedding costs of the second sub-image to asymmetric, facilitating the secure embedding of the remaining secret information. To support our phased embedding approach, our GAN's discriminator incorporates two steganalyers with different tasks: distinguishing the generator's final output, i.e., the stego image, from both the input cover image and the partially embedded stego image, providing diverse guidance to the generator. In addition, we introduce a simple yet effective update strategy to ensure a stable training process. Comprehensive experiments demonstrate that our method significantly enhances security over existing symmetric steganography techniques, achieving state-of-the-art levels compared to other methods focused on embedding costs adjustments. Additionally, detailed ablation studies validate our approach's effectiveness. © 2024 ACM.

关键词： Steganography

来源：评论

学校读者我要写书评

暂无评论

HairDiffusion: Vivid Multi-Colored Hair Editing via Latent Diffusion 38

HairDiffusion: Vivid Multi-Colored Hair Editing via Latent D...

引用

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Zeng, Yu Zhang, Yang Liu, Jiachen Shen, Linlin Deng, Kaijun He, Weizhao Wang, Jinbao Computer Vision Institute School of Computer Science & Software Engineering Shenzhen University China Shenzhen Institute of Artificial Intelligence and Robotics for Society China National Engineering Laboratory for Big Data System Computing Technology Shenzhen University China Guangdong Provincial Key Laboratory of Intelligent Information Processing China

Hair editing is a critical image synthesis task that aims to edit hair color and hairstyle using text descriptions or reference images, while preserving irrelevant attributes (e.g., identity, background, cloth). Many existing methods are based on StyleGAN to address this task. However, due to the limited spatial distribution of StyleGAN, it struggles with multiple hair color editing and facial preservation. Considering the advancements in diffusion models, we utilize Latent Diffusion Models (LDMs) for hairstyle editing. Our approach introduces Multi-stage Hairstyle Blend (MHB), effectively separating control of hair color and hairstyle in diffusion latent space. Additionally, we train a warping module to align the hair color with the target region. To further enhance multi-color hairstyle editing, we fine-tuned a CLIP model using a multi-color hairstyle dataset. Our method not only tackles the complexity of multi-color hairstyles but also addresses the challenge of preserving original colors during diffusion editing. Extensive experiments showcase the superiority of our method in editing multi-color hairstyles while preserving facial attributes given textual descriptions and reference images. © 2024 Neural information processing systems foundation. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Knowledge-Guided Cross-Topic Visual Question Generation 30

Knowledge-Guided Cross-Topic Visual Question Generation

引用

Joint 30th International Conference on Computational Linguistics and 14th International Conference on Language Resources and Evaluation, LREC-COLING 2024

作者： Liu, Hongfei Wang, Guohua Xie, Jiayuan Chen, Jiali Fang, Wenhao Cai, Yi School of Software Engineering South China University of Technology China Key Laboratory of Big Data and Intelligent Robot South China University of Technology Ministry of Education China Department of Computing The Hong Kong Polytechnic University Hong Kong

ISBN: (纸本)9782493814104

Visual question generation (VQG) task aims to generate high-quality questions based on the input image. Current methods primarily focus on generating questions containing specified content utilizing answers or question types as constraints. However, these constraints make it challenging to control the topic of generated questions (e.g., conversation or test subject topics) for various applications. Thus, it is necessary to utilize topics as constraints to guide question generation. Considering that there are many topics and it is almost impossible for human annotations to cover them, we propose the cross-topic learning VQG (CTL-VQG) task, which aims to generate questions related to unseen topics in cross-topic scenarios. In this paper, we propose a knowledge-guided cross-topic visual question generation (KC-VQG) model to extract unseen topic-related information for question generation. Specifically, an image-topic feature extractor is introduced in our model to extract topic-related intuitive visual features;an image-topic knowledge extractor is used to extract and select the most appropriate topic-related implicit knowledge from large language models for generating questions. Extensive experiments show that our model outperforms baselines and can effectively generate unseen topic-related questions in cross-topic scenarios. © 2024 ELRA Language Resource Association: CC BY-NC 4.0.

关键词： Data mining

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：