检索结果-内蒙古大学图书馆

Augmenting Context Representation with Triggers Knowledge for Relation Extraction 12th

学校读者我要写书评

暂无评论

Augmenting Context Representation with Triggers Knowledge fo...

12th IFIP TC 12 International Conference on Intelligent Information Processing, IIP 2022

作者： Li, En Shi, Shumin Yang, Zhikun Huang, He Yan School of Computer Science and Technology Beijing Institute of Technology Beijing China Beijing Engineering Research Center of High Volume Language Information Processing and Cloud Computing Applications Beijing China

ISBN: (纸本)9783031039478

Relation Extraction (RE) requires the model to classify the correct relation from a set of relation candidates given the corresponding sentence and two entities. Recent work mainly studies how to utilize more data or incorporate extra context information especially with Pre-trained Language Models (PLMs). However, these models still face with the challenges of avoiding being affected by irrelevant or misleading words. In this paper, we propose a novel model to help alleviate such deficiency. Specifically, our model automatically mines the triggers of the sentence iteratively with the sentence itself from the previous iteration, and augment the semantics of the context representation from BERT with both entity pair and triggers skillfully. We conduct extensive experiments to evaluate the proposed model and effectively obtain empirical improvement in TACRED. © 2022, IFIP International Federation for Information Processing.

关键词： Semantics

IMPROVING COMPLEX REASONING WITH DYNAMIC PROMPT CORRUPTION: A SOFT PROMPT OPTIMIZATION APPROACH

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Fan, Sinan Xie, Liang Shen, Chen Teng, Ge Yuan, Xiaosong Zhang, Xiaofeng Huang, Chenxi Wang, Wenxiao He, Xiaofei Ye, Jieping Zhejiang University China Hangzhou YunQi Academy of Engineering China College of Computer Science and Technology Zhejiang University of Technology China Alibaba Cloud Computing China

Prompt-tuning (PT) for large language models (LLMs) can facilitate the performance on various conventional NLP tasks with significantly fewer trainable parameters. However, our investigation reveals that PT provides limited improvement and may even degrade the primitive performance of LLMs on complex reasoning tasks. Such a phenomenon suggests that soft prompts can positively impact certain instances while negatively affecting others, particularly during the later phases of reasoning. To address these challenges, We first identify an information accumulation within the soft prompts. Through detailed analysis, we demonstrate that this phenomenon is often accompanied by erroneous information flow patterns in the deeper layers of the model, which ultimately lead to incorrect reasoning outcomes. we propose a novel method called Dynamic Prompt Corruption (DPC) to take better advantage of soft prompts in complex reasoning tasks, which dynamically adjusts the influence of soft prompts based on their impact on the reasoning process. Specifically, DPC consists of two stages: Dynamic Trigger and Dynamic Corruption. First, Dynamic Trigger measures the impact of soft prompts, identifying whether beneficial or detrimental. Then, Dynamic Corruption mitigates the negative effects of soft prompts by selectively masking key tokens that interfere with the reasoning process. We validate the proposed approach through extensive experiments on various LLMs and reasoning tasks, including GSM8K, MATH, and AQuA. Experimental results demonstrate that DPC can consistently enhance the performance of PT, achieving 4%-8% accuracy gains compared to vanilla prompt tuning, highlighting the effectiveness of our approach and its potential to enhance complex reasoning in LLMs. © 2025, CC BY.

关键词： Modeling languages

Blend the Separated: Mixture of Synergistic Experts for Data-Scarcity Drug-Target Interaction Prediction

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Zhai, Xinlong Wang, Chunchen Wang, Ruijia Kang, Jiazheng Li, Shujie Chen, Boyu Ma, Tengfei Zhou, Zikai Yang, Cheng Shi, Chuan Beijing University of Posts and Telecommunications China China Telecom Cloud Computing Research Institute China College of Computer Science and Electronic Engineering Hunan University China

Drug-target interaction prediction (DTI) is essential in various applications including drug discovery and clinical application. There are two perspectives of input data widely used in DTI prediction: Intrinsic data represents how drugs or targets are constructed, and extrinsic data represents how drugs or targets are related to other biological entities. However, any of the two perspectives of input data can be scarce for some drugs or targets, especially for those unpopular or newly discovered. Furthermore, ground-truth labels for specific interaction types can also be scarce. Therefore, we propose the first method to tackle DTI prediction under input data and/or label scarcity. To make our model functional when only one perspective of input data is available, we design two separate experts to process intrinsic and extrinsic data respectively and fuse them adaptively according to different samples. Furthermore, to make the two perspectives complement each other and remedy label scarcity, two experts synergize with each other in a mutually supervised way to exploit the enormous unlabeled data. Extensive experiments on 3 real-world datasets under different extents of input data scarcity and/or label scarcity demonstrate our model outperforms states of the art significantly and steadily, with a maximum improvement of 53.53%. We also test our model without any data scarcity and it still outperforms current methods. © 2025, CC BY-NC-ND.

关键词： Prediction models

Blend the Separated: Mixture of Synergistic Experts for Data-Scarcity Drug-Target Interaction Prediction 39

学校读者我要写书评

暂无评论

Blend the Separated: Mixture of Synergistic Experts for Data...

39th Annual AAAI Conference on Artificial Intelligence, AAAI 2025

ISBN: (纸本)157735897X

关键词：

Towards Robust Video Anomaly Detection in 6G Networks: A Scene-Adaptive Framework

学校读者我要写书评

暂无评论

IEEE Network 2025年

作者： Zhang, Sijia Yu, Yong Tao, Xiaoling Wang, Lianhai Shaanxi Normal University School of Computer Science China Guilin University of Electronic Technology Guangxi Cooperative Innovation Center of Cloud Computing and Big Data China Qilu University of Technology Shandong Academy of Sciences Shandong Provincial Key Laboratory of Computing Power Internet and Service Computing China

With the rapid advancement of 5th Generation (5G) networks, their low latency and high bandwidth facilitate the development of smart city systems. Among the smart city systems, anomaly detection in surveillance videos has been an essential part due to its importance to public safety and security. With the transition to 6G technology, the practical applications of video anomaly detection are expanding to encompass a variety of scenarios, such as sidewalks, shopping malls and campuses, which exhibit significant differences. However, most existing anomaly detection methods deployed in the cloud typically rely on the fixed datasets for training and apply the trained uniform model to the edge services across different scenarios. The distinct environmental conditions may lead to varying performance of a uniform anomaly detection model. To alleviate these issues, we propose a scene-adaptive video anomaly detection framework in 6G-based applications. The proposed framework leverages the advantages of 6G networks to facilitate localized processing and rapid adaptation to diverse environments. To enable the anomaly detection model to adapt more robustly to local environments, the proposed framework integrates a pre-trained model and simulates anomalies to train the anomaly detection model in a denoising manner. By integrating pre-trained teacher models with edge computing nodes, the proposed method facilitates accurate anomaly detection while maintaining robust performance across varying scenarios. The framework utilizes both original data and simulated anomalous data to fine-tune the edge model, ensuring it aligns with the unique characteristics of each environment. Experimental results demonstrate the effectiveness of the proposed method across different scenarios, enabling smarter and more responsive surveillance systems in the 6G era. © 2025 IEEE.

关键词： Video analysis

Eventually-Consistent Federated Scheduling for Data Center Workloads

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Thiyyakat, Meghana Kalambur, Subramaniam Chaudhary, Rishit Nayak, Saurav G. Shetty, Adarsh Sitaram, Dinkar Department of Computer Science and Engineering PES University Bangalore India Cloud Computing Innovation Council of India Bangalore India

Data center schedulers operate at unprecedented scales today to accommodate the growing demand for computing and storage power. The challenge that schedulers face is meeting the requirements of scheduling speeds despite the scale. To do so, most scheduler architectures use parallelism. However, these architectures consist of multiple parallel scheduling entities that can only utilize partial knowledge of the data center’s state, as maintaining consistent global knowledge or state would involve considerable communication overhead. The disadvantage of scheduling without global knowledge is sub-optimal placements—tasks may be made to wait in queues even though there are resources available in zones outside the scope of the scheduling entity’s state. This leads to unnecessary queuing overheads and lower resource utilization of the data center. In this paper, extend our previous work on Megha, a federated decentralized data center scheduling architecture that uses eventual consistency. The architecture utilizes both parallelism and an eventually-consistent global state in each of its scheduling entities to make fast decisions in a scalable manner. In our work, we compare Megha with 3 scheduling architectures – Sparrow, Eagle, and Pigeon, using simulation. We also evaluate Megha’s prototype on a 123-node cluster and compare its performance with Pigeon’s prototype using cluster traces. The results of our experiments show that Megha consistently reduces delays in job completion time when compared to other architectures. © 2023, CC BY.

关键词： Digital storage

Structural and Statistical Texture Knowledge Distillation for Semantic Segmentation

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Ji, Deyi Wang, Haoran Tao, Mingyuan Huang, Jianqiang Hua, Xian-Sheng Lu, Hongtao Alibaba Cloud Computing Ltd. China Department of Computer Science and Engineering Shanghai Jiao Tong University China

Existing knowledge distillation works for semantic segmentation mainly focus on transferring high-level contextual knowledge from teacher to student. However, low-level texture knowledge is also of vital importance for characterizing the local structural pattern and global statistical property, such as boundary, smoothness, regularity and color contrast, which may not be well addressed by high-level deep features. In this paper, we are intended to take full advantage of both structural and statistical texture knowledge and propose a novel Structural and Statistical Texture Knowledge Distillation (SSTKD) framework for semantic segmentation. Specifically, for structural texture knowledge, we introduce a Contourlet Decomposition Module (CDM) that decomposes low-level features with iterative Laplacian pyramid and directional filter bank to mine the structural texture knowledge. For statistical knowledge, we propose a Denoised Texture Intensity Equalization Module (DTIEM) to adaptively extract and enhance statistical texture knowledge through heuristics iterative quantization and denoised operation. Finally, each knowledge learning is supervised by an individual loss function, forcing the student network to mimic the teacher better from a broader perspective. Experiments show that the proposed method achieves state-of-the-art performance on Cityscapes, Pascal VOC 2012 and ADE20K datasets. Copyright © 2023, The Authors. All rights reserved.

关键词： Textures

Analytical Study of the Features and Parameters of Off-Angle Iris Image vs. Deep Learning Model

学校读者我要写书评

暂无评论

Analytical Study of the Features and Parameters of Off-Angle...

2024 International Conference on Decision Aid sciences and Applications, DASA 2024

作者： Suleiman, Aghiad Khedr Shafek, Darin Mahfuri, Mahmoud Khishe, Mohammad Saeed, Ali Ghwanmeh, Sameh Alma'Moon University College Cybersecurity and Cloud Computing Technologies Engineering Dept. Baghdad Iraq Alma'Moon University College Computer Engineering Techniques Dept. Baghdad Iraq International University for Science and Technology Computer and Information Technology Department Syria Innovation Center for Artificial Intelligence Applications Yuan Ze University Taiwan Jadara Research Center Jadara University Jordan Jordan Computer Center Northern Technical University Ninevah Iraq College of Computer Information Technology Computer Science Dept American University in the Emirates United Arab Emirates

ISBN: (纸本)9798350369106

This paper presents an analytical study and a technique for extracting the features of a common case of images of the iris called off-angle iris which was taken for persons identification systems. The main problem when using biological iris measurements to identify the persons is the difficulty of identifying and extracting features of the iris. This problem is increasing when dealing with off-angle iris and it leads to decreased system accuracy and increased system rate error. In return, all the transfer learning techniques face difficulties in the case of heavily degraded data and partially occluded, and off-angle. A new method has a whole new methodology to deal with the image as it is without transformation processes. A new algorithm has been included for extracting features of the iris through the pupil switching points. The most discriminating points of the iris depend on the biological human eye statistics and analytical study. It has been trained and tested on the common images of the off-angle iris database so-called: 'CASIA Iris 1.0'. It has been implemented in the MATLAB environment. The results showed the efficiency of this technique, high precision and more importantly low failed acceptance rate. It emphasized that it is adaptive as well as efficiency improvement of the system. © 2024 IEEE.

关键词： Contrastive Learning

QoS-Aware Decentralized Trustworthy Forensic Evidence Management Framework Using the Internet of Vehicle Things (IoVT) 3rd

学校读者我要写书评

暂无评论

QoS-Aware Decentralized Trustworthy Forensic Evidence Manage...

3rd International Conference on Advanced computing and Applications, ICACA 2024

作者： Das, Puja Singh, Moutushi Roy, Deepsubhra Guha Department of Computer Science and Business Systems KPRIET Uthupalayam641407 India Department of Information Technology Institute of Engineering and Management IEM Centre of Excellence for InnovAI University of Engineering and Management Kolkata India Institute of Engineering and Management IEM Centre of Excellence for Cloud Computing and IoT University of Engineering and Management Kolkata700091 India

ISBN: (纸本)9789819747986

Vehicles are becoming physical computer systems that connect with other cars and collect data from numerous pieces of equipment. These improvements contribute to developing smart and connected automobiles that will supply valuable data to owners, producers, insurance firms, and repair network operators for various uses. In post-accident circumstances, especially for auto vehicles (self-driving cars), using transportation data might help identify the guilty party. Verification, accessibility, secrecy, and dependability are all privacy needs for a vehicular ad hoc network. These criteria are linked to critical attacks that make operating a network difficult. However, to access different types of information in automobiles, we proposed a permissioned blockchain framework to organize the acquired transportation information among the numerous parts concerned. We incorporate the public key exchange protocol on the vehicle into the projected blockchain-based forensic system to ensure participation identification and anonymity. The proposed system offers a deep learning-based predictive incident modeling with blockchain for the incident data stored in the blockchain. Moreover, several methods have been combined in forensic management with blockchain to establish an auto-evaluation scheme. As a result, we extensively evaluated, analyzed, and discussed the works and contributions of skilled researchers to uncover new avenues in an ongoing and prospective study. The main goal of this research might be to explore the future potential for smart and connected automobiles using blockchain. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Vehicular ad hoc networks