Abstract:This paper designs a dialogue model based on the granulated knowledge graph (Granulation Knowledge aware Dialogue Model, GKDM). Through the method of structural granulation, the knowledge graph is divided into two modules: local knowledge graph and non-local knowledge graph, respectively capturing the local semantics adjacent to the knowledge entity and the overall semantic representation of the located knowledge sub-graph, which solves the problem of the division of the knowledge graph. This paper designs the static graph attention mechanism and the hierarchical dynamic graph attention mechanism to participate in the process of encoding and decoding in the model respectively, and aggregates the semantics of knowledge entities through multi-level attention weights, and fuses them into the text vector, which solves the problem of the fusion of multi-layer knowledge information. Based on the above two innovations, the model proposed in this paper alleviates the problems of the current knowledge graph-based dialogue model that cannot capture the reasoning meaning of the multi-hop paths of the knowledge graph and the inconsistent theme of the reply context.

Alert Name:
Alerting to:
Authentication email will be sent to your email address in 24 hours
Frequency:
Email Message Format:	Plain text Graphical(HTML)

Complete the form below and we will recommend the selected titles to your friends on your behalf. * Indicates a required field.
Your name*:
Your email address*:
Recipient's name*:
Recipient's email address*:
(multiple recipient's names and email addresses should be separated with semicolons)
Your comments:	I thought you would find the page(s) useful.

Your name:
Your email address:
Recipient's name:
Recipient's email address:
(multiple recipient's names and email addresses should be separated with semicolons)
Your comments:	I thought you would find this page useful.

Disclaimer: This message was sent to your friend using the "Send it to a friend" facility on the Sciencepaper Online’ WWW site, http://www.paper.edu.cn/en. The Sciencepaper Online is not responsible for the content of this email, and anything said in this email does not necessarily reflect the Sciencepaper Online's views.


	1. Granularized Dialogue Model of The Knowledge Graph
	Zhao Meiyong,Yu Wen
	Computer Science and Technology 10 April 2024
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (0 B)

	Abstract:This paper designs a dialogue model based on the granulated knowledge graph (Granulation Knowledge aware Dialogue Model, GKDM). Through the method of structural granulation, the knowledge graph is divided into two modules: local knowledge graph and non-local knowledge graph, respectively capturing the local semantics adjacent to the knowledge entity and the overall semantic representation of the located knowledge sub-graph, which solves the problem of the division of the knowledge graph. This paper designs the static graph attention mechanism and the hierarchical dynamic graph attention mechanism to participate in the process of encoding and decoding in the model respectively, and aggregates the semantics of knowledge entities through multi-level attention weights, and fuses them into the text vector, which solves the problem of the fusion of multi-layer knowledge information. Based on the above two innovations, the model proposed in this paper alleviates the problems of the current knowledge graph-based dialogue model that cannot capture the reasoning meaning of the multi-hop paths of the knowledge graph and the inconsistent theme of the reply context.
	TO cite this article:Zhao Meiyong,Yu Wen. Granularized Dialogue Model of The Knowledge Graph[OL].[10 April 2024] http://en.paper.edu.cn/en_releasepaper/content/4763289


	2. Multi-proposal Collaboration and Multi-task Training for Weakly-supervised Video Moment Retrieval
	ZHANG Bolin,YANG Chao,JIANG Bin,KOMAMIZU Takahiro,IDE Ichiro
	Computer Science and Technology 02 April 2024
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (0 B)

	Abstract:This study focuses on weakly-supervised Video Moment Retrieval (VMR), aiming to identify a moment semantically similar to the given query within an untrimmed video using only video-level correspondences, without relying on temporal annotations during training. Previous methods either aggregate predictions for all instances in the video, or indirectly address the task by proposing reconstructions for the query. However, these methods often produce low-quality temporal proposals, struggle with distinguishing misaligned moments in the same video, or lack stability due to a reliance on a single auxiliary task.To address these limitations, we present a novel weakly-supervised method called Multi-proposal Collaboration and Multi-task Training (MCMT). Initially, we generate multiple proposals and derive corresponding learnable Gaussian masks from them. These masks are then combined to create a high-quality positive sample mask, highlighting video clips most relevant to the query. Concurrently, we classify other clips in the same video as easy negative sample and the entire video as hard negative sample. During training, we introduce forward and inverse masked query reconstruction tasks to impose more substantial constraints on the network, promoting more robust and stable retrieval performance. Extensive experiments on two standard benchmarks affirm the effectiveness of the proposed method in VMR.
	TO cite this article:ZHANG Bolin,YANG Chao,JIANG Bin, et al. Multi-proposal Collaboration and Multi-task Training for Weakly-supervised Video Moment Retrieval[OL].[ 2 April 2024] http://en.paper.edu.cn/en_releasepaper/content/4763143

Saved Papers

Saved Papers


	3. The Analysis of Adversarial Robustness in Federated Learning Based on Differential Privacy
	Li Qi,Li Li
	Computer Science and Technology 26 March 2024
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (0 B)

	Abstract:Federated learning (FL) allows multiple participants to train models collaboratively by keeping their data local while exchanging updates, however, this also increases the risk of attack. Models in FL are as vulnerable as centrally trained models against adversarial examples. And with frequent data exchanging, the risk of data breaches increases. To enhance the robustness against adversarial examples, this paper introduces the differential privacy into FL, which is also an effective mechanism used for privacy protection.Many popular approaches that involve adversarial training in FL often compromise test accuracy while effect. In this work, a novel method is proposed which applies differential privacy to enhance both adversarial robustness and privacy protection level. Extensive experiments on multiple datasets demonstrate the proposed approach has a better performance compared with other baselines.
	TO cite this article:Li Qi,Li Li. The Analysis of Adversarial Robustness in Federated Learning Based on Differential Privacy[OL].[26 March 2024] http://en.paper.edu.cn/en_releasepaper/content/4763096


	4. Multi-Object Tracking with decoupled representations and unreliable detections in complex scenes
	XIAO Yuan, ZOU Qi
	Computer Science and Technology 21 March 2024
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (0 B)

	Abstract:Joint-Detection-and-Embedding paradigm achieves fast tracking by simultaneously learning detection and Re-ID features. However, it still faces performance degradation in complex scenes and the misalignment between detection and Re-ID features. In this paper, we propose a decoupling module based on channel-wise attention mechanism to obtain task-aligned features served for different demands of detection and Re-ID. To improve the performance of data association, we fuse motion, location, appearance information and perform a two-round matching for high and low confidence detections respectively by the Motion-GIoU matrix and the Embedding-GIoU matrix. Additionally, we apply the camera motion compensation to get a more accurate motion estimation, resulting in a more robust tracking in the scenes of camera motion and low-frame-rate. Extensive experiments show that our proposed method outperforms a wide range of existing methods on the MOTChallenge and HiEvE datasets.
	TO cite this article:XIAO Yuan, ZOU Qi. Multi-Object Tracking with decoupled representations and unreliable detections in complex scenes[OL].[21 March 2024] http://en.paper.edu.cn/en_releasepaper/content/4763002


	5. End-to-End Virtual Shadow Generation Based on Shadow Detection
	Xue Junsheng, Huang Hai
	Computer Science and Technology 21 March 2024
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (0 B)

	Abstract:With the rapid development of the augmented reality field, virtual shadow generation technology has garnered widespread attention. However, traditional methods involve complex computational requirements for environmental modeling and display, and the effectiveness of using fewer 3D parameter estimation methods is suboptimal. To address these issues, this paper proposes an end-to-end virtual shadow generation method based on shadow detection. Firstly, we introduce a shadow detector designed to identify real shadows and their corresponding occlusions in the background context, while also learning the mapping relationship between occlusions and shadows in the current scene. Secondly, we devise a virtual shadow generator. Utilizing the mapping relationship obtained from the detector as guidance information, it is encoded into the generator's input. Through feature extraction, encoding, and decoding of the input information, we ultimately obtain the virtual shadow image. Experimental results demonstrate the exceptional performance of the proposed virtual shadow generation method. In comparison to existing methods for direct virtual shadow generation, our approach significantly enhances the harmony and realism of the synthesized images.
	TO cite this article:Xue Junsheng, Huang Hai. End-to-End Virtual Shadow Generation Based on Shadow Detection[OL].[21 March 2024] http://en.paper.edu.cn/en_releasepaper/content/4762999


	6. Causal-Proto SSRL: Learning Dynamic-Necessary State Variables for Multi-Environment Reinforcement Learning
	Zhang Meng,Zhang Chunhong,Hu Zheng,Zhuang Benhui
	Computer Science and Technology 19 March 2024
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (0 B)

	Abstract:The ability to learn directly from high-dimensional observations such as pixels allows reinforcement learning(RL) to achieve more widely applications. However, high-dimensional observations contain entangled task-relevant and task-irrelevant informations, as well as informations related to actions but unnecessary, which leads to non-essential dependencies and thus affects the generalization and robustness of the reinforcement learning. In order to learn abstract state representations from high-dimensional observations which generalized acorss multiple tasks and robust in environments with different task-unnecessary informations, this paper formulates the POMDP as Partially Observable Temporal Causal Dynamic Models (POTCDMs) and proposes a self-supervised RL with causal representation learning, Causal-Proto RL. This method seperates encoded observations into dynamic-necessary and dynamic-unnecessary state variables where only dynamic-necessary state variables are fed into RL by predicting the causal relationships simultaneously. This method is pretrained in absence of specific task rewards with an intrinsic rewards fo curiosity of causal relationships and are implemented in multiple difficult downstream tasks. This paper evaluate the algorithm in DeepMind Control Suit. This algorithm performs as well as other SOTA slef-supervised RL on a series of downstream tasks in environents as same as pretraining, and demonstrates generalization and robustness in downstream task environments different from pretraining.
	TO cite this article:Zhang Meng,Zhang Chunhong,Hu Zheng, et al. Causal-Proto SSRL: Learning Dynamic-Necessary State Variables for Multi-Environment Reinforcement Learning[OL].[19 March 2024] http://en.paper.edu.cn/en_releasepaper/content/4762946


	7. Research on the hippocampus medical imaging segmentation method for small samples
	QI Shu-Wen,Jiang Zhu-qing1,Jiang Zhu-qing1,Jiang Zhu-qing1
	Computer Science and Technology 16 March 2024
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (0 B)

	Abstract:The hippocampus is located between the thalamus and the medial temporal lobe. It is mainly responsible for cognition, learning, and long and short memory. It is closely related to many diseases such as Alzheimer's disease and temporal lobe epilepsy. Therefore, the accurate segmentation of the hippocampal structure in magnetic resonance imaging is of great significance for the diagnosis of brain injury and brain disease prediction in clinical medicine. In recent years, the rapid development of deep learning technology has brought about brand-new changes to the field of hippocampal segmentation. Deep learning is data-driven, and the quantity and quality of data directly affect the accuracy of hippocampal segmentation. However, due to the difficulty of MR imaging acquisition and expensive manual annotation, hippocampus MR imaging is relatively scarce, which limits the performance improvement of deep learning models in hippocampal segmentation tasks to some extent. In order to overcome the challenges in small sample data scenarios and improve the accuracy of hippocampal segmentation, this paper proposes a data augmentation method, which aims to expand the data (brain magnetic resonance images) and label (hippocampus mask) simultaneously, so as to alleviate the problem of data scarcity and annotation scarcity. Through experiments, the proposed method can effectively improve the accuracy of hippocampal segmentation.
	TO cite this article:QI Shu-Wen,Jiang Zhu-qing1,Jiang Zhu-qing1, et al. Research on the hippocampus medical imaging segmentation method for small samples[OL].[16 March 2024] http://en.paper.edu.cn/en_releasepaper/content/4762832


	8. TransFuseNet: A Novel Multi-task Model for Community-Acquired Pneumonia Segmentation and Classification
	CHE PeiShuai,YIN Si-Xing,LI Shu-Fang
	Computer Science and Technology 13 March 2024
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (0 B)

	Abstract:Community-acquired pneumonia (CAP) poses a global public health challenge, and in the current environment of the pneumonia pandemic, timely and accurate diagnosis of different types of pneumonia is particularly crucial. Computed Tomography (CT) is an effective means of diagnosing pneumonia, and the use of artificial intelligence (AI) for diagnostic assistance can enhance clinical diagnostic efficiency. Therefore, this paper introduces a 3D multitask deep learning model called TransFuseNet to achieve real-time and accurate segmentation and classification of CAP.Specifically, the proposed network consists of two sub-networks: a 3D scSEU-Net sub-network for pneumonia lesion segmentation and a classification sub-network based on a fully convolutional Transformer. Both sub-networks share the same encoder, where the segmentation branch captures local features and spatial relationships, while the classification branch performs long-range modeling to capture global context information. Simultaneously, a loss function is introduced to enhance the interaction between the two sub-networks, balancing the importance of the two tasks.The retrospective dataset includes 180 patients who underwent thin-slice chest CT scans at a medical center in China. Numerous experiments demonstrate that the model achieved AUC: 0.989, DSC: 0.723, average accuracy: 0.927, precision: 0.889, sensitivity: 0.866, and specificity: 0.835 on the test set. The model shows no significant difference in pneumonia detection accuracy compared to radiologists.
	TO cite this article:CHE PeiShuai,YIN Si-Xing,LI Shu-Fang. TransFuseNet: A Novel Multi-task Model for Community-Acquired Pneumonia Segmentation and Classification[OL].[13 March 2024] http://en.paper.edu.cn/en_releasepaper/content/4762404


	9. A Multi-Document Inference Method Based on TR-BERT and Attention Networks
	ZHAO Jiaqi,LIN Rongheng
	Computer Science and Technology 13 March 2024
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (0 B)

	Abstract:In the task of multi-document inference, information related to the answer may reside across multiple relevant texts, and sometimes this information is not directly associated with the question. To address the challenge of balancing accuracy and efficiency in multi-document inference for the electric utility customer service scenario, this paper proposes a multi-level inference method based on pre-trained models and attention networks. The model utilizes pre-trained models to preserve the rich semantics extracted from paragraph texts and questions, and then evaluates the relevance of candidates through an attention mechanism. Furthermore, due to the real-time requirements of the question-answering scenario, we employ the TR-BERT pre-trained model based on dynamic token reduction and simplify the attention network. Experimental results on the WikiHop dataset demonstrate that the model overall exhibits advantages in both computational speed and accuracy, providing effective methodological support for the multi-turn question-answering functionality in intelligent question-answering systems.
	TO cite this article:ZHAO Jiaqi,LIN Rongheng. A Multi-Document Inference Method Based on TR-BERT and Attention Networks[OL].[13 March 2024] http://en.paper.edu.cn/en_releasepaper/content/4762558


	10. Named Entity Recognition in the Perovskite Field Based on Convolutional Neural Networks and MatBERT
	ZHANG Jiaxin,ZHANG Lingxue,SUN Yuxuan,LI Wei,QUHE Ruge
	Computer Science and Technology 13 March 2024
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (0 B)

	Abstract:Due to the significant increase in publications in the field of materials science, there has been a bottleneck in organizing material science knowledge and discovering new materials. The number of literature in the emerging field of perovskite materials has grown to a massive scale. It is necessary to compile information on the structure, properties, synthesis methods, characterization techniques, and applications of perovskite materials. To address this issue, we employ named entity recognition, a natural language processing technique, to extract important entities from perovskite material texts. In this paper, we propose a method based on convolutional neural networks (CNN) and MatBERT. Firstly, we utilize MatBERT, which has been pre-trained on a large amount of material science text, to generate contextualized word embeddings. Next, we extract feature information using a CNN model. Finally, a conditional random field (CRF) layer is used for decoding sequences in addition to calculating the training and validation loss. Experimental results demonstrate that the performance of our model on perovskite material dataset is improved by 1%~6% compared with BERT, SciBERT and MatBERT models. Through this model, we extract the entities of 2389 abstracts to obtain knowledge of perovskite materials.
	TO cite this article:ZHANG Jiaxin,ZHANG Lingxue,SUN Yuxuan, et al. Named Entity Recognition in the Perovskite Field Based on Convolutional Neural Networks and MatBERT[OL].[13 March 2024] http://en.paper.edu.cn/en_releasepaper/content/4762696

	Check out RSS, or use RSS reader to subscribe this item