Tsinghua Science and Technology最新文献

{"title":"Power of Second Opportunity: Dynamic Pricing with Second Chance","authors":"Chensheng Ma;Shaojie Tang;Zhao Zhang","doi":"10.26599/TST.2023.9010108","DOIUrl":"https://doi.org/10.26599/TST.2023.9010108","url":null,"abstract":"In this paper, we consider the following dynamic pricing problem. Suppose the market price \u0000<tex>$v_{t}$</tex>\u0000 of an item arriving at time \u0000<tex>$t$</tex>\u0000 is determined by \u0000<tex>$v_{t}=pmb{theta}^{mathrm{T}}pmb{x}_{t}$</tex>\u0000, where \u0000<tex>$pmb{x}_{t}$</tex>\u0000 is the feature vector of that item and \u0000<tex>$pmb{theta}$</tex>\u0000 is an unknown vector parameter. The seller has to post prices without knowing \u0000<tex>$pmb{theta}$</tex>\u0000 such that the total regret in time span \u0000<tex>$T$</tex>\u0000 is minimized. Considering real-world scenarios in which people may negotiate prices, we propose a model called Second Chance Pricing, in which a seller has a second opportunity to post a price after the first offer is declined. Theoretical analysis shows that a second chance of pricing results in a total regret between \u0000<tex>$o(frac{ln T}{nln n}+frac{1}{n})$</tex>\u0000 and \u0000<tex>$O(n^{2}ln T)$</tex>\u0000, where \u0000<tex>$n$</tex>\u0000 is the dimension of the feature space. Experiments on both synthetic data and real data demonstrate significant benefits brought about by the second chance where the regret is only 13% of that of one chance.","PeriodicalId":48690,"journal":{"name":"Tsinghua Science and Technology","volume":"30 2","pages":"543-560"},"PeriodicalIF":6.6,"publicationDate":"2024-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10786952","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142825972","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Optimizing Data Distributions Based on Jensen-Shannon Divergence for Federated Learning","authors":"Zhiyao Hu;Dongsheng Li;Ke Yang;Ying Xu;Baoyun Peng","doi":"10.26599/TST.2023.9010091","DOIUrl":"https://doi.org/10.26599/TST.2023.9010091","url":null,"abstract":"In current federated learning frameworks, a central server randomly selects a small number of clients to train local models at the beginning of each global iteration. Since clients' local data are non-dependent and identically distributed, partial local models are not consistent with the global model. Existing studies employ model cleaning methods to find inconsistent local models. Model cleaning methods measure the cosine similarity between local models and the global model. The inconsistent local model is cleaned out and will not be aggregated for the next global model. However, model cleaning methods incur negative effects such as large computation overheads and limited updates. In this paper, we propose a data distribution optimization method, called federated distribution optimization (FedDO), aiming to overcome the shortcomings of model cleaning methods. FedDO calculates the gradient of the Jensen-Shannon divergence to decrease the discrepancy between selected clients' data distribution and the overall data distribution. We test our method on the multi-classification regression model, the multi-layer perceptron, and the convolutional neural network model on a handwritten digital image dataset. Compared with model cleaning methods, FedDO improves the training accuracy by 1.8%, 2.6%, and 5.6%, respectively.","PeriodicalId":48690,"journal":{"name":"Tsinghua Science and Technology","volume":"30 2","pages":"670-681"},"PeriodicalIF":6.6,"publicationDate":"2024-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10786940","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142825786","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Layered Temporal Spatial Graph Attention Reinforcement Learning for Multiplex Networked Industrial Chains Energy Management","authors":"Yuanshuang Jiang;Kai Di;Xingyu Wu;Zhongjian Hu;Fulin Chen;Pan Li;Yichuan Jiang","doi":"10.26599/TST.2023.9010111","DOIUrl":"https://doi.org/10.26599/TST.2023.9010111","url":null,"abstract":"Demand response has recently become an essential means for businesses to reduce production costs in industrial chains. Meanwhile, the current industrial chain structure has also become increasingly complex, forming new characteristics of multiplex networked industrial chains. Fluctuations in real-time electricity prices in demand response propagate through the coupling and cascading relationships within and among these network layers, resulting in negative impacts on the overall energy management cost. However, existing demand response methods based on reinforcement learning typically focus only on individual agents without considering the influence of dynamic factors on intra and inter-network relationships. This paper proposes a Layered Temporal Spatial Graph Attention (LTSGA) reinforcement learning algorithm suitable for demand response in multiplex networked industrial chains to address this issue. The algorithm first uses Long Short-Term Memory (LSTM) to learn the dynamic temporal characteristics of electricity prices for decision-making. Then, LTSGA incorporates a layered spatial graph attention model to evaluate the impact of dynamic factors on the complex multiplex networked industrial chain structure. Experiments demonstrate that the proposed LTSGA approach effectively characterizes the influence of dynamic factors on intra- and inter-network relationships within the multiplex industrial chain, enhancing convergence speed and algorithm performance compared with existing state-of-the-art algorithms.","PeriodicalId":48690,"journal":{"name":"Tsinghua Science and Technology","volume":"30 2","pages":"528-542"},"PeriodicalIF":6.6,"publicationDate":"2024-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10786953","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142825866","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Detection of Acute Lymphoblastic Leukemia Using a Novel Bone Marrow Image Segmentation","authors":"M. Anline Rejula;Ben M JebIn;Ravi Selvakumar;S. Amutha;Eberlein George","doi":"10.26599/TST.2023.9010099","DOIUrl":"https://doi.org/10.26599/TST.2023.9010099","url":null,"abstract":"In our study, we present a novel method for automating the segmentation and classification of bone marrow images to distinguish between normal and Acute Lymphoblastic Leukaemia (ALL). Built upon existing segmentation techniques, our approach enhances the dual threshold segmentation process, optimizing the isolation of nucleus and cytoplasm components. This is achieved by adapting threshold values based on image characteristics, resulting in superior segmentation outcomes compared to previous methods. To address challenges, such as noise and incomplete white blood cells, we employ mathematical morphology and median filtering techniques. These methods effectively denoise the images and remove incomplete cells, leading to cleaner and more precise segmentation. Additionally, we propose a unique feature extraction method using a hybrid discrete wavelet transform, capturing both spatial and frequency information. This allows for the extraction of highly discriminative features from segmented images, enhancing the reliability of classification. For classification purposes, we utilize an improved Adaptive Neuro-Fuzzy Inference System (ANFIS) that leverages the extracted features. Our enhanced classification algorithm surpasses traditional methods, ensuring accurate identification of acute lymphoblastic leukaemia. Our innovation lies in the comprehensive integration of segmentation techniques, advanced denoising methods, novel feature extraction, and improved classification. Through extensive evaluation on bone marrow samples from the Acute Lymphoblastic Leukemia Image DataBase (ALL-IDB) for Image Processing database using MATLAB 10.0, our method demonstrates outstanding classification accuracy. The segmentation accuracy for various cell types, including Band cells (96%), Metamyelocyte (99%), Myeloblast (96%), N. myelocyte (97%), N. promyelocyte (97%), and Neutrophil cells (98%), further underscores the potential of our approach as a high-quality tool for ALL diagnosis.","PeriodicalId":48690,"journal":{"name":"Tsinghua Science and Technology","volume":"30 2","pages":"610-623"},"PeriodicalIF":6.6,"publicationDate":"2024-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10786930","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142825973","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A Survey of the Application of Neural Networks to Event Extraction","authors":"Jianye Xie;Yulan Zhang;Huaizhen Kou;Xiaoran Zhao;Zhikang Feng;Lekang Song;Weiyi Zhong","doi":"10.26599/TST.2023.9010139","DOIUrl":"https://doi.org/10.26599/TST.2023.9010139","url":null,"abstract":"Event extraction is an important part of natural language information extraction, and it's widely employed in other natural language processing tasks including question answering and machine reading comprehension. However, there is a lack of recent comprehensive survey papers on event extraction. In the past few years, numerous high-quality and innovative event extraction methods have been proposed, making it necessary to consolidate these new developments with previous work in order to provide a clear overview for researchers and serve as a reference for future studies. In addition, event detection is a fundamental sub-task in event extraction, previous survey papers have often overlooked the related work on event detection. Therefore, this paper aims to bridge these gaps by presenting a comprehensive survey of event extraction, including recent advancements and an analysis of previous research on event detection. The resources for event extraction are first introduced in this research, and then the numerous neural network models currently employed in event extraction tasks are divided into four types: word sequence-based methods, graph-based neural network methods, external knowledge-based approaches, and prompt-based approaches. We compare and contrast them in depth, pointing out the flaws and difficulties with existing research. Finally, we discuss the future of event extraction development.","PeriodicalId":48690,"journal":{"name":"Tsinghua Science and Technology","volume":"30 2","pages":"748-768"},"PeriodicalIF":6.6,"publicationDate":"2024-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10786941","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142825915","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Diversity-Based Recruitment in Crowdsensing by Combinatorial Multi-Armed Bandits","authors":"Abdalaziz Sawwan;Jie Wu","doi":"10.26599/TST.2024.9010053","DOIUrl":"https://doi.org/10.26599/TST.2024.9010053","url":null,"abstract":"Mobile Crowdsensing (MCS) represents a transformative approach to collecting data from the environment as it utilizes the ubiquity and sensory capabilities of mobile devices with human participants. This paradigm enables scales of data collection critical for applications ranging from environmental monitoring to urban planning. However, the effective harnessing of this distributed data collection capability faces significant challenges. One of the most significant challenges is the variability in the sensing qualities of the participating devices while they are initially unknown and must be learned over time to optimize task assignments. This paper tackles the dual challenges of managing task diversity to mitigate data redundancy and optimizing task assignment amidst the inherent variability of worker performance. We introduce a novel model that dynamically adjusts task weights based on assignment frequency to promote diversity and incorporates a flexible approach to account for the different qualities of task completion, especially in scenarios with overlapping task assignments. Our strategy aims to maximize the overall weighted quality of data collected within the constraints of a predefined budget. Our strategy leverages a combinatorial multi-armed bandit framework with an upper confidence bound approach to guide decision-making. We demonstrate the efficacy of our approach through a combination of regret analysis and simulations grounded in realistic scenarios.","PeriodicalId":48690,"journal":{"name":"Tsinghua Science and Technology","volume":"30 2","pages":"732-747"},"PeriodicalIF":6.6,"publicationDate":"2024-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10786946","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142825914","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Joint Extraction of Uygur Medicine Knowledge with Edge Computing","authors":"Fan Lu;Quan Qi;Huaibin Qin","doi":"10.26599/TST.2024.9010006","DOIUrl":"https://doi.org/10.26599/TST.2024.9010006","url":null,"abstract":"Edge computing, a novel paradigm for performing computations at the network edge, holds significant relevance in the healthcare domain for extracting medical knowledge from traditional Uygur medical texts. Medical knowledge extraction methods based on edge computing deploy deep learning models on edge devices to achieve localized entity and relation extraction. This approach avoids transferring substantial sensitive data to cloud data centers, effectively safeguarding the privacy of healthcare services. However, existing relation extraction methods mainly employ a sequential pipeline approach, which classifies relations between determined entities after entity recognition. This mode faces challenges such as error propagation between tasks, insufficient consideration of dependencies between the two subtasks, and the neglect of interrelations between different relations within a sentence. To address these challenges, a joint extraction model with parameter sharing in edge computing is proposed, named CoEx-Bert. This model leverages shared parameterization between two models to jointly extract entities and relations. Specifically, CoEx-Bert employs two models, each separately sharing hidden layer parameters, and combines these two loss functions for joint backpropagation to optimize the model parameters. Additionally, it effectively resolves the issue of entity overlapping when extracting knowledge from unstructured Uygur medical texts by considering contextual relations. Finally, this model is deployed on edge devices for real-time extraction and inference of Uygur medical knowledge. Experimental results demonstrate that CoEx-Bert outperforms existing state-of-the-art methods, achieving accuracy, recall, and F1-score of 90.65%, 92.45%, and 91.54%, respectively, in the Uygur traditional medical literature dataset. These improvements represent a 6.45% increase in accuracy, a 9.45% increase in recall, and a 7.95% increase in F1-score compared to the baseline.","PeriodicalId":48690,"journal":{"name":"Tsinghua Science and Technology","volume":"30 2","pages":"782-795"},"PeriodicalIF":6.6,"publicationDate":"2024-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10786944","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142825864","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Maximization of k-Submodular Function with d-Knapsack Constraints Over Sliding Window","authors":"Wenqi Wang;Yuefang Sun;Zhiren Sun;Donglei Du;Xiaoyan Zhang","doi":"10.26599/TST.2023.9010121","DOIUrl":"https://doi.org/10.26599/TST.2023.9010121","url":null,"abstract":"Submodular function maximization problem has been extensively studied recently. A natural variant of submodular function is k-submodular function, which has many applications in real life, such as influence maximization and sensor placement problem. The domain of a \u0000<tex>$k$</tex>\u0000 -submodular function has \u0000<tex>$k$</tex>\u0000 disjoint subsets, and hence includes submodular function as a special case when \u0000<tex>$k=1$</tex>\u0000. This work investigates the k-submodular function maximization problem with d-knapsack constraints over the sliding window. Based on the smooth histogram technique, we design a deterministic approximation algorithm. Furthermore, we propose a randomized algorithm to improve the approximation ratio.","PeriodicalId":48690,"journal":{"name":"Tsinghua Science and Technology","volume":"30 2","pages":"488-498"},"PeriodicalIF":6.6,"publicationDate":"2024-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10786950","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142825913","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Key Mechanisms on Resource Optimization Allocation in Minority Game Based on Reinforcement Learning","authors":"Changyan Di;Tianyi Wang;Qingguo Zhou;Jinqiang Wang","doi":"10.26599/TST.2023.9010155","DOIUrl":"https://doi.org/10.26599/TST.2023.9010155","url":null,"abstract":"The emergence of coordinated and consistent macro behavior among self-interested individuals competing for limited resources represents a central inquiry in comprehending market mechanisms and collective behavior. Traditional economics tackles this challenge through a mathematical and theoretical lens, assuming individuals are entirely rational and markets tend to stabilize through the price mechanism. Our paper addresses this issue from an econophysics standpoint, employing reinforcement learning to construct a multi-agent system modeled on minority games. Our study has undertaken a comparative analysis from both collective and individual perspectives, affirming the pivotal roles of reward feedback and individual memory in addressing the aforementioned challenge. Reward feedback serves as the guiding force for the evolution of collective behavior, propelling it towards an overall increase in rewards. Individuals, drawing insights from their own rewards through accumulated learning, gain information about the collective state and adjust their behavior accordingly. Furthermore, we apply information theory to present a formalized equation for the evolution of collective behavior. Our research supplements existing conclusions regarding the mechanisms of a free market and, at a micro level, unveils the dynamic evolution of individual behavior in synchronization with the collective.","PeriodicalId":48690,"journal":{"name":"Tsinghua Science and Technology","volume":"30 2","pages":"721-731"},"PeriodicalIF":6.6,"publicationDate":"2024-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10786935","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142825787","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Study of Driver's Perception in Driving Tasks Based on Naturalistic Driving Experiments and fNIRS Measurement","authors":"Bilu Li;Xin Pei;Dan Zhang;Xinmiao Zhang;Zhuoran Li;Duanrui Yu;Shifei Shen","doi":"10.26599/TST.2024.9010002","DOIUrl":"https://doi.org/10.26599/TST.2024.9010002","url":null,"abstract":"Understanding how drivers perceive and respond to external stimuli in driving tasks is important for the development of advanced driving technologies and human-computer interaction. In this paper, we conducted a temporal response analysis between driving data and cortical activation data measured by functional near-infrared spectroscopy (fNIRS), based on a naturalistic driving experiment. Temporal response function analysis indicates that stimuli, which elicit significant responses of drivers include distance, acceleration, time headway, and the velocity of the preceding vehicle. For these stimuli, the time lags and response patterns were further discussed. The influencing factors on drivers' perception were also studied based on various driver characteristics. These conclusions can provide guidance for the construction of carfollowing models, the safety assessment of drivers and the improvement of advanced driving technologies.","PeriodicalId":48690,"journal":{"name":"Tsinghua Science and Technology","volume":"30 2","pages":"796-812"},"PeriodicalIF":6.6,"publicationDate":"2024-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10786947","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142797890","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}