Background:N-acetyltransferase 10(NAT10)was reported to be associated with the immune microenvironment in several cancers.However,it is not known in pancreatic ductal adenocarcinoma(PDAC).This study aimed to elucidate...Background:N-acetyltransferase 10(NAT10)was reported to be associated with the immune microenvironment in several cancers.However,it is not known in pancreatic ductal adenocarcinoma(PDAC).This study aimed to elucidate the roles and mechanisms of NAT10 in tumor malignancy and the tumor microenvironment(TME)in PDAC.Methods:NAT10 expression and its role in tumor progression and clinical prognosis were analyzed using bioinformatics and functional assays.Downstream genes regulated by NAT10 and their underlying mechanisms were explored using acetylated RNA immunoprecipitation,quantitative polymerase chain reaction,RNA immunoprecipitation,and Western blotting.The role and mechanism of NAT10 in the PDAC TME were further explored using bioinformatics,single-cell RNA sequencing,multiplexed immunofluorescence,and flow cytometry.The association between NAT10 and immunotherapeutic response was investigated in a mouse model by inhibiting the programmed cell death 1/programmed cell death ligand 1(PD-1/PD-L1)axis with a PD-1/PD-L1 binding inhibitor,Naamidine J.Results:NAT10 was upregulated in PDAC tissues and cell lines,and was associated with poor progression-free survival of PDAC patients.NAT10 promoted tumor progression by enhancing the mRNA stability of lamininβ3(LAMB3)via N4-acetylation modification,thereby activating the focal adhesion kinase(FAK)/extracellular regulated protein kinases(ERK)pathway.NAT10 promoted subcutaneous tumor growth,increased the proportion of exhausted CD8+T cells(CD8+Tex),especially the intermediate CD8+Tex subset,and decreased the proportion of cytotoxic CD8+T cell(CD8+Tc)subset in the PDAC TME.Naamidine J treatment significantly enhanced the proportion of CD8+Tc subset and reduced the proportion of intermediate CD8+Tex subset in mice bearing subcutaneous tumors with high NAT10 expression.Regarding the regulatory mechanism,NAT10 increased PD-L1 expression and abundance in tumor cells by activating the LAMB3/FAK/ERK pathway,thereby reducing the cytotoxicity of CD8+T cells.Inhibition of the PD-1/PD-L1 axis with Naamidine J retrieved CD8+T cell cytotoxicity.Conclusions:This study proposes a regulatory role of NAT10 in tumor progression and immune microenvironment via the LAMB3/FAK/ERK pathway in PDAC.These findings may favor the selection of candidates who may benefit from immunotherapy,optimize current therapeutic strategies,and improve the clinical prognosis of PDAC patients.展开更多
Time series classification is related to many dif- ferent domains, such as health informatics, finance, and bioinformatics. Due to its broad applications, researchers have developed many algorithms for this kind of ta...Time series classification is related to many dif- ferent domains, such as health informatics, finance, and bioinformatics. Due to its broad applications, researchers have developed many algorithms for this kind of tasks, e.g., multivariate time series classification. Among the classifi- cation algorithms, k-nearest neighbor （k-NN） classification （particularly 1-NN） combined with dynamic time warping （DTW） achieves the state of the art performance. The defi- ciency is that when the data set grows large, the time con- sumption of 1-NN with DTW will be very expensive. In con- trast to 1-NN with DTW, it is more efficient but less ef- fective for feature-based classification methods since their performance usually depends on the quality of hand-crafted features. In this paper, we aim to improve the performance of traditional feature-based approaches through the feature learning techniques. Specifically, we propose a novel deep learning framework, multi-channels deep convolutional neu- ral networks （MC-DCNN）, for multivariate time series classi- fication. This model first learns features from individual uni- variate time series in each channel, and combines information from all channels as feature representation at the final layer. Then, the learnt features are applied into a multilayer percep- tron （MLP） for classification. Finally, the extensive experi- ments on real-world data sets show that our model is not only more efficient than the state of the art but also competitive in accuracy. This study implies that feature learning is worth to be investigated for the problem of time series classification.展开更多
Information Extraction(IE)aims to extract structural knowledge from plain natural language texts.Recently,generative Large Language Models(LLMs)have demonstrated remarkable capabilities in text understanding and gener...Information Extraction(IE)aims to extract structural knowledge from plain natural language texts.Recently,generative Large Language Models(LLMs)have demonstrated remarkable capabilities in text understanding and generation.As a result,numerous works have been proposed to integrate LLMs for IE tasks based on a generative paradigm.To conduct a comprehensive systematic review and exploration of LLM efforts for IE tasks,in this study,we survey the most recent advancements in this field.We first present an extensive overview by categorizing these works in terms of various IE subtasks and techniques,and then we empirically analyze the most advanced methods and discover the emerging trend of IE tasks with LLMs.Based on a thorough review conducted,we identify several insights in technique and promising research directions that deserve further exploration in future studies.We maintain a public repository and consistently update related works and resources on GitHub(LLM4IE repository).展开更多
Recently,many online Karaoke(KTV)platforms have been released,where music lovers sing songs on these platforms.In the meantime,the system automatically evaluates user proficiency according to their singing behavior.Re...Recently,many online Karaoke(KTV)platforms have been released,where music lovers sing songs on these platforms.In the meantime,the system automatically evaluates user proficiency according to their singing behavior.Recommending approximate songs to users can initialize singers5 participation and improve users,loyalty to these platforms.However,this is not an easy task due to the unique characteristics of these platforms.First,since users may be not achieving high scores evaluated by the system on their favorite songs,how to balance user preferences with user proficiency on singing for song recommendation is still open.Second,the sparsity of the user-song interaction behavior may greatly impact the recommendation task.To solve the above two challenges,in this paper,we propose an informationfused song recommendation model by considering the unique characteristics of the singing data.Specifically,we first devise a pseudo-rating matrix by combing users’singing behavior and the system evaluations,thus users'preferences and proficiency are leveraged.Then we mitigate the data sparsity problem by fusing users*and songs'rich information in the matrix factorization process of the pseudo-rating matrix.Finally,extensive experimental results on a real-world dataset show the effectiveness of our proposed model.展开更多
Distributed stochastic gradient descent and its variants have been widely adopted in the training of machine learning models,which apply multiple workers in parallel.Among them,local-based algorithms,including Local S...Distributed stochastic gradient descent and its variants have been widely adopted in the training of machine learning models,which apply multiple workers in parallel.Among them,local-based algorithms,including Local SGD and FedAvg,have gained much attention due to their superior properties,such as low communication cost and privacypreserving.Nevertheless,when the data distribution on workers is non-identical,local-based algorithms would encounter a significant degradation in the convergence rate.In this paper,we propose Variance Reduced Local SGD(VRL-SGD)to deal with the heterogeneous data.Without extra communication cost,VRL-SGD can reduce the gradient variance among workers caused by the heterogeneous data,and thus it prevents local-based algorithms from slow convergence rate.Moreover,we present VRL-SGD-W with an effectivewarm-up mechanism for the scenarios,where the data among workers are quite diverse.Benefiting from eliminating the impact of such heterogeneous data,we theoretically prove that VRL-SGD achieves a linear iteration speedup with lower communication complexity even if workers access non-identical datasets.We conduct experiments on three machine learning tasks.The experimental results demonstrate that VRL-SGD performs impressively better than Local SGD for the heterogeneous data and VRL-SGD-W is much robust under high data variance among workers.展开更多
The propagation of information in online social networks plays a critical role in modern life,and thus has been studied broadly.Researchers have proposed a series of propagation models,generally,which use a single tra...The propagation of information in online social networks plays a critical role in modern life,and thus has been studied broadly.Researchers have proposed a series of propagation models,generally,which use a single transition probability or consider factors such as content and time to describe the way how a user activates her/his neighbors.However,the research on the mechanism how social ties between users play roles in propagation process is still limited.Specifically,comprehensive summary of factors which affect user’s decision whether to share neighbor’s content was lacked in existing works,so that the existing models failed to clearly describe the process a user be activated by a neighbor.To this end,in this paper,we analyze the close correspondence between social tie in propagation process and communication channel,thus we propose to exploit the communication channel to describe the information propagation process between users,and design a social tie channel(STC)model.The model can naturally incorporate many factors affecting the information propagation through edges such as content topic and user preference,and thus can effectively capture the user behavior and relationship characteristics which indicate the property of a social tie.Extensive experiments conducted on two real-world datasets demonstrate the effectiveness of our model on content sharing prediction between users.展开更多
Adaptive learning rate methods have been successfully applied in many fields,especially in training deep neural networks.Recent results have shown that adaptive methods with exponential increasing weights on squared p...Adaptive learning rate methods have been successfully applied in many fields,especially in training deep neural networks.Recent results have shown that adaptive methods with exponential increasing weights on squared past gradients(i.e.,ADAM,RMSPROP)may fail to converge to the optimal solution.Though many algorithms,such as AMSGRAD and ADAMNC,have been proposed to fix the non-convergence issues,achieving a data-dependent regret bound similar to or better than ADAGRAD is still a challenge to these methods.In this paper,we propose a novel adaptive method weighted adaptive algorithm(WADA)to tackle the non-convergence issues.Unlike AMSGRAD and ADAMNC,we consider using a milder growing weighting strategy on squared past gradient,in which weights grow linearly.Based on this idea,we propose weighted adaptive gradient method framework(WAGMF)and implement WADA algorithm on this framework.Moreover,we prove that WADA can achieve a weighted data-dependent regret bound,which could be better than the original regret bound of ADAGRAD when the gradients decrease rapidly.This bound may partially explain the good performance of ADAM in practice.Finally,extensive experiments demonstrate the effectiveness of WADA and its variants in comparison with several variants of ADAM on training convex problems and deep neural networks.展开更多
The use of traditional positioning technologies, such as GPS and wireless local positioning, rely on un- derlying infrastructure. However, in a subway environment, such positioning systems are not available for the po...The use of traditional positioning technologies, such as GPS and wireless local positioning, rely on un- derlying infrastructure. However, in a subway environment, such positioning systems are not available for the position- ing tasks, such as the detection of the train arrivals for the passengers in the train. An alternative approach is to exploit the contextual information available in the mobile devices of subway riders to detect train arrivals. To this end, we pro- pose to exploit multiple contextual features extracted from the mobile devices of subway riders to precisely detecting train arrivals. Following this line, we first investigate poten- tial contextual features which may be effective to detect train arrivals according to the observations from 3D accelerome- ters and GSM radio. Furthermore, we propose to explore the maximum entropy （MaxEnt） model for training a train ar- rival detector by learning the correlation between contextual features and train arrivals. Finally, we perform extensive ex- periments on several real-world data sets collected from two major subway lines in the Beijing subway system. Experi- mental results validate both the effectiveness and efficiency of the proposed approach.展开更多
Network embedding,which targets at learning the vector representation of vertices,has become a crucial issue in network analysis.However,considering the complex structures and heterogeneous attributes in real-world ne...Network embedding,which targets at learning the vector representation of vertices,has become a crucial issue in network analysis.However,considering the complex structures and heterogeneous attributes in real-world networks,existing methods may fail to handle the inconsistencies between the structure topology and attribute proximity.Thus,more comprehensive techniques are urgently required to capture the highly non-linear network structure and solve the existing inconsistencies with retaining more information.To that end,in this paper,we propose a heterogeneous-attributes enhancement deep framework(HEDF),which could better capture the non-linear structure and associated information in a deep learningway,and effectively combine the structure information of multi-views by the combining layer.Along this line,the inconsistencies will be handled to some extent and more structure information will be preserved through a semi-supervised mode.The extensive validations on several real-world datasets show that our model could outperform the baselines,especially for the sparse and inconsistent situation with less training data.展开更多
Recently advancements in deep learning models have significantly facilitated the development of sequential recommender systems(SRS).However,the current deep model structures are limited in their ability to learn high-...Recently advancements in deep learning models have significantly facilitated the development of sequential recommender systems(SRS).However,the current deep model structures are limited in their ability to learn high-quality embeddings with insufficient data.Meanwhile,highly skewed long-tail distribution is very common in recommender systems.Therefore,in this paper,we focus on enhancing the representation of tail items to improve sequential recommendation performance.Through empirical studies on benchmarks,we surprisingly observe that both the ranking performance and training procedure are greatly hindered by the poorly optimized tail item embeddings.To address this issue,we propose a sequential recommendation framework named TailRec that enables contextual information of tail item well-leveraged and greatly improves its corresponding representation.Given the characteristics of the sequential recommendation task,the surrounding interaction records of each tail item are regarded as contextual information without leveraging any additional side information.This approach allows for the mining of contextual information from cross-sequence behaviors to boost the performance of sequential recommendations.Such a light contextual filtering component is plug-and-play for a series of SRS models.To verify the effectiveness of the proposed TailRec,we conduct extensive experiments over several popular benchmark recommenders.The experimental results demonstrate that TailRec can greatly improve the recommendation results and speed up the training process.The codes of our methods have been available.展开更多
Vehicle Color Recognition(VCR)plays a vital role in intelligent traffic management and criminal investigation assistance.However,the existing vehicle color datasets only cover 13 classes,which can not meet the current...Vehicle Color Recognition(VCR)plays a vital role in intelligent traffic management and criminal investigation assistance.However,the existing vehicle color datasets only cover 13 classes,which can not meet the current actual demand.Besides,although lots of efforts are devoted to VCR,they suffer from the problem of class imbalance in datasets.To address these challenges,in this paper,we propose a novel VCR method based on Smooth Modulation Neural Network with Multi-Scale Feature Fusion(SMNN-MSFF).Specifically,to construct the benchmark of model training and evaluation,we first present a new VCR dataset with 24 vehicle classes,Vehicle Color-24,consisting of 10091 vehicle images from a 100-hour urban road surveillance video.Then,to tackle the problem of long-tail distribution and improve the recognition performance,we propose the SMNN-MSFF model with multiscale feature fusion and smooth modulation.The former aims to extract feature information from local to global,and the latter could increase the loss of the images of tail class instances for training with class-imbalance.Finally,comprehensive experimental evaluation on Vehicle Color-24 and previously three representative datasets demonstrate that our proposed SMNN-MSFF outperformed state-of-the-art VCR methods.And extensive ablation studies also demonstrate that each module of our method is effective,especially,the smooth modulation efficiently help feature learning of the minority or tail classes.Vehicle Color-24 and the code of SMNN-MSFF are publicly available and can contact the author to obtain.展开更多
With the increasing frequency of natural disasters and health emergencies,wearable infrared thermal imaging devices are becoming more prevalent in fire protection and medical fields.However,these devices often face im...With the increasing frequency of natural disasters and health emergencies,wearable infrared thermal imaging devices are becoming more prevalent in fire protection and medical fields.However,these devices often face imaging performance challenges such as insufficient contrast,dark areas and blurred edges,which significantly limit their practical effectiveness.To tackle these challenges,we propose a novel unsupervised lightweight 3D convolutional network(UL3DCN)specifically designed for enhancing infrared images on wearable devices.In this framework,the task of infrared image enhancement is conceptualized as generating high dynamic range infrared images from the corresponding temperature sequences during thermal equilibrium.To achieve this,we first design a learnable dynamic filtering module tailored for simulating a series of infrared image sequences under varying temperature differences.This module extends a single image from the spatial domain into the spatio-temporal domain.Subsequently,we employ a lightweight 3D convolution module to effectively extract spatio-temporal information from the image sequence.Finally,inspired by Zero-DCE,we utilize the extracted information to estimate pixel values and highorder curves,thereby enhancing the infrared images.Comprehensive experimental results demonstrate that our method achieves outstanding performance and real-time capabilities.Additionally,the proposed UL3DCN model has been successfully integrated into a wearable infrared firefighting mask.展开更多
Recently,with the rapid advancements in Large Language Models(LLMs),LLM-based Open-domain Question Answering(OpenQA)methods have reaped the benefits of emergent understanding and answering capabilities enabled by mass...Recently,with the rapid advancements in Large Language Models(LLMs),LLM-based Open-domain Question Answering(OpenQA)methods have reaped the benefits of emergent understanding and answering capabilities enabled by massive parameters compared to traditional methods.However,most of these methods encounter two critical challenges:how to integrate knowledge into LLMs effectively and how to adaptively generate results with specific answer formats.To address these challenges,we propose a novel framework,which aims to improve the OpenQA performance by exploring knowledge integration and controllable generation on LLMs simultaneously,namely GenKI.Specifically,we first train a dense passage retrieval model to retrieve associated knowledge from a given knowledge base.Subsequently,we introduce a novel knowledge integration model that incorporates the retrieval knowledge into instructions during fine-tuning to intensify the model.Furthermore,to enable controllable generation in LLMs,we leverage a certain fine-tuned LLM and an ensemble framework based on text consistency incorporating all coherence,fluency,and answer format assurance.Finally,extensive experiments conducted on three datasets with diverse answer formats demonstrate the effectiveness of GenKI with comparison of state-of-the-art baselines.Moreover,ablation studies have disclosed a linear relationship between the frequency of retrieved knowledge and the model’s ability to recall knowledge accurately with the ground truth.Tests focusing on the out-of-domain scenario and knowledge base independence scenario have further affirmed the robustness and controllable capability of GenKI.Our code of GenKI is available at https://github.com/USTC-StarTeam/GenKI.展开更多
基金supported by the National Natural Science Foundation of China(1127105011371183+2 种基金61403036)the Science and Technology Development Foundation of CAEP(2013A04030202013B0403068)
基金Jiangsu 1184 CHEN et al.Natural Science Foundation Project(No.BK20191142)Scientific Research Project of Jiangsu Provincial Health Commission(No.M2024075)+2 种基金High-level talent training project of Wuxi Taihu Talent Plan(No.BJ2020055,BJ2023048)Clinical Medical Research Project of the Affiliated Hospital of Jiangnan University(No.LCYJ202321)Postgraduate Research&Practice Innovation Program of Jiangsu Province(No.KYCX24_2648).
文摘Background:N-acetyltransferase 10(NAT10)was reported to be associated with the immune microenvironment in several cancers.However,it is not known in pancreatic ductal adenocarcinoma(PDAC).This study aimed to elucidate the roles and mechanisms of NAT10 in tumor malignancy and the tumor microenvironment(TME)in PDAC.Methods:NAT10 expression and its role in tumor progression and clinical prognosis were analyzed using bioinformatics and functional assays.Downstream genes regulated by NAT10 and their underlying mechanisms were explored using acetylated RNA immunoprecipitation,quantitative polymerase chain reaction,RNA immunoprecipitation,and Western blotting.The role and mechanism of NAT10 in the PDAC TME were further explored using bioinformatics,single-cell RNA sequencing,multiplexed immunofluorescence,and flow cytometry.The association between NAT10 and immunotherapeutic response was investigated in a mouse model by inhibiting the programmed cell death 1/programmed cell death ligand 1(PD-1/PD-L1)axis with a PD-1/PD-L1 binding inhibitor,Naamidine J.Results:NAT10 was upregulated in PDAC tissues and cell lines,and was associated with poor progression-free survival of PDAC patients.NAT10 promoted tumor progression by enhancing the mRNA stability of lamininβ3(LAMB3)via N4-acetylation modification,thereby activating the focal adhesion kinase(FAK)/extracellular regulated protein kinases(ERK)pathway.NAT10 promoted subcutaneous tumor growth,increased the proportion of exhausted CD8+T cells(CD8+Tex),especially the intermediate CD8+Tex subset,and decreased the proportion of cytotoxic CD8+T cell(CD8+Tc)subset in the PDAC TME.Naamidine J treatment significantly enhanced the proportion of CD8+Tc subset and reduced the proportion of intermediate CD8+Tex subset in mice bearing subcutaneous tumors with high NAT10 expression.Regarding the regulatory mechanism,NAT10 increased PD-L1 expression and abundance in tumor cells by activating the LAMB3/FAK/ERK pathway,thereby reducing the cytotoxicity of CD8+T cells.Inhibition of the PD-1/PD-L1 axis with Naamidine J retrieved CD8+T cell cytotoxicity.Conclusions:This study proposes a regulatory role of NAT10 in tumor progression and immune microenvironment via the LAMB3/FAK/ERK pathway in PDAC.These findings may favor the selection of candidates who may benefit from immunotherapy,optimize current therapeutic strategies,and improve the clinical prognosis of PDAC patients.
文摘Time series classification is related to many dif- ferent domains, such as health informatics, finance, and bioinformatics. Due to its broad applications, researchers have developed many algorithms for this kind of tasks, e.g., multivariate time series classification. Among the classifi- cation algorithms, k-nearest neighbor （k-NN） classification （particularly 1-NN） combined with dynamic time warping （DTW） achieves the state of the art performance. The defi- ciency is that when the data set grows large, the time con- sumption of 1-NN with DTW will be very expensive. In con- trast to 1-NN with DTW, it is more efficient but less ef- fective for feature-based classification methods since their performance usually depends on the quality of hand-crafted features. In this paper, we aim to improve the performance of traditional feature-based approaches through the feature learning techniques. Specifically, we propose a novel deep learning framework, multi-channels deep convolutional neu- ral networks （MC-DCNN）, for multivariate time series classi- fication. This model first learns features from individual uni- variate time series in each channel, and combines information from all channels as feature representation at the final layer. Then, the learnt features are applied into a multilayer percep- tron （MLP） for classification. Finally, the extensive experi- ments on real-world data sets show that our model is not only more efficient than the state of the art but also competitive in accuracy. This study implies that feature learning is worth to be investigated for the problem of time series classification.
基金supported in part by the grants from the National Natural Science Foundation of China(Nos.62222213,62072423)partially supported by Research Impact Fund(No.R1015-23),APRC-CityU New Research Initiatives(No.9610565,Start-up Grant for New Faculty of CityU)+7 种基金CityU-HKIDS Early Career Research Grant(No.9360163)Hong Kong ITC Innovation and Technology Fund Midstream Research Programme for Universities Project(No.ITS/034/22MS)Hong Kong Environmental and Conservation Fund(No.88/2022)SIRG-CityU Strategic Interdisciplinary Research Grant(No.7020046)Huawei(Huawei Innovation Research Program),Tencent(CCFTencent Open Fund,Tencent Rhino-Bird Focused Research Program),Ant Group(CCF-Ant Research Fund,Ant Group Research Fund)Alibaba(CCFAlimama Tech Kangaroo Fund(No.2024002))CCF-BaiChuan-Ebtech Foundation Model FundKuaishou.
文摘Information Extraction(IE)aims to extract structural knowledge from plain natural language texts.Recently,generative Large Language Models(LLMs)have demonstrated remarkable capabilities in text understanding and generation.As a result,numerous works have been proposed to integrate LLMs for IE tasks based on a generative paradigm.To conduct a comprehensive systematic review and exploration of LLM efforts for IE tasks,in this study,we survey the most recent advancements in this field.We first present an extensive overview by categorizing these works in terms of various IE subtasks and techniques,and then we empirically analyze the most advanced methods and discover the emerging trend of IE tasks with LLMs.Based on a thorough review conducted,we identify several insights in technique and promising research directions that deserve further exploration in future studies.We maintain a public repository and consistently update related works and resources on GitHub(LLM4IE repository).
基金grants from the National Key Research and Development Program of China(2016YFB1000904)the National Natural Science Foundation of China(Grant Nos.61325010 and U 1605251)+3 种基金the Fundamental Research Funds for the Central Universities of China(WK2350000001)Le Wu gratefully acknowledges the support of the Open Project Program of the National Laboratory of Pattern Recognition(201700017)the Fundamental Research Funds for the Central Universities(JZ2016HGBZ0749)Yong Ge acknowledges the support of the National Natural Science Foundation of China(NSFC,Grant Nos.61602234 and 61572032).
文摘Recently,many online Karaoke(KTV)platforms have been released,where music lovers sing songs on these platforms.In the meantime,the system automatically evaluates user proficiency according to their singing behavior.Recommending approximate songs to users can initialize singers5 participation and improve users,loyalty to these platforms.However,this is not an easy task due to the unique characteristics of these platforms.First,since users may be not achieving high scores evaluated by the system on their favorite songs,how to balance user preferences with user proficiency on singing for song recommendation is still open.Second,the sparsity of the user-song interaction behavior may greatly impact the recommendation task.To solve the above two challenges,in this paper,we propose an informationfused song recommendation model by considering the unique characteristics of the singing data.Specifically,we first devise a pseudo-rating matrix by combing users’singing behavior and the system evaluations,thus users'preferences and proficiency are leveraged.Then we mitigate the data sparsity problem by fusing users*and songs'rich information in the matrix factorization process of the pseudo-rating matrix.Finally,extensive experimental results on a real-world dataset show the effectiveness of our proposed model.
基金This research was partially supported by grants from the National Key Research and Development Program of China(No.2018YFC0832101)the National Natural Science Foundation of China(Grant Nos.U20A20229 and 61922073).
文摘Distributed stochastic gradient descent and its variants have been widely adopted in the training of machine learning models,which apply multiple workers in parallel.Among them,local-based algorithms,including Local SGD and FedAvg,have gained much attention due to their superior properties,such as low communication cost and privacypreserving.Nevertheless,when the data distribution on workers is non-identical,local-based algorithms would encounter a significant degradation in the convergence rate.In this paper,we propose Variance Reduced Local SGD(VRL-SGD)to deal with the heterogeneous data.Without extra communication cost,VRL-SGD can reduce the gradient variance among workers caused by the heterogeneous data,and thus it prevents local-based algorithms from slow convergence rate.Moreover,we present VRL-SGD-W with an effectivewarm-up mechanism for the scenarios,where the data among workers are quite diverse.Benefiting from eliminating the impact of such heterogeneous data,we theoretically prove that VRL-SGD achieves a linear iteration speedup with lower communication complexity even if workers access non-identical datasets.We conduct experiments on three machine learning tasks.The experimental results demonstrate that VRL-SGD performs impressively better than Local SGD for the heterogeneous data and VRL-SGD-W is much robust under high data variance among workers.
基金supported by the National Natural Science Foundation of China(Grants Nos.U1605251,61727809 and 91546110)the Youth Innovation Promotion Association of CAS(2014299)Special Program for Applied Research on Super Computation of the NSFCGuangdong Joint Fund(the second phase).
文摘The propagation of information in online social networks plays a critical role in modern life,and thus has been studied broadly.Researchers have proposed a series of propagation models,generally,which use a single transition probability or consider factors such as content and time to describe the way how a user activates her/his neighbors.However,the research on the mechanism how social ties between users play roles in propagation process is still limited.Specifically,comprehensive summary of factors which affect user’s decision whether to share neighbor’s content was lacked in existing works,so that the existing models failed to clearly describe the process a user be activated by a neighbor.To this end,in this paper,we analyze the close correspondence between social tie in propagation process and communication channel,thus we propose to exploit the communication channel to describe the information propagation process between users,and design a social tie channel(STC)model.The model can naturally incorporate many factors affecting the information propagation through edges such as content topic and user preference,and thus can effectively capture the user behavior and relationship characteristics which indicate the property of a social tie.Extensive experiments conducted on two real-world datasets demonstrate the effectiveness of our model on content sharing prediction between users.
基金We thank the anonymous reviewers for their insightful comments and discussions.This research was partially supported by grants from the National Key Research and Development Program of China(2018YFB1004300)the National Natural Science Foundation of China(Grant Nos.61703386,61727809,and U1605251).
文摘Adaptive learning rate methods have been successfully applied in many fields,especially in training deep neural networks.Recent results have shown that adaptive methods with exponential increasing weights on squared past gradients(i.e.,ADAM,RMSPROP)may fail to converge to the optimal solution.Though many algorithms,such as AMSGRAD and ADAMNC,have been proposed to fix the non-convergence issues,achieving a data-dependent regret bound similar to or better than ADAGRAD is still a challenge to these methods.In this paper,we propose a novel adaptive method weighted adaptive algorithm(WADA)to tackle the non-convergence issues.Unlike AMSGRAD and ADAMNC,we consider using a milder growing weighting strategy on squared past gradient,in which weights grow linearly.Based on this idea,we propose weighted adaptive gradient method framework(WAGMF)and implement WADA algorithm on this framework.Moreover,we prove that WADA can achieve a weighted data-dependent regret bound,which could be better than the original regret bound of ADAGRAD when the gradients decrease rapidly.This bound may partially explain the good performance of ADAM in practice.Finally,extensive experiments demonstrate the effectiveness of WADA and its variants in comparison with several variants of ADAM on training convex problems and deep neural networks.
文摘The use of traditional positioning technologies, such as GPS and wireless local positioning, rely on un- derlying infrastructure. However, in a subway environment, such positioning systems are not available for the position- ing tasks, such as the detection of the train arrivals for the passengers in the train. An alternative approach is to exploit the contextual information available in the mobile devices of subway riders to detect train arrivals. To this end, we pro- pose to exploit multiple contextual features extracted from the mobile devices of subway riders to precisely detecting train arrivals. Following this line, we first investigate poten- tial contextual features which may be effective to detect train arrivals according to the observations from 3D accelerome- ters and GSM radio. Furthermore, we propose to explore the maximum entropy （MaxEnt） model for training a train ar- rival detector by learning the correlation between contextual features and train arrivals. Finally, we perform extensive ex- periments on several real-world data sets collected from two major subway lines in the Beijing subway system. Experi- mental results validate both the effectiveness and efficiency of the proposed approach.
基金This research was partially supported by the National Natural Science Foundation of China(Grants Nos.U1605251 and 61727809).
文摘Network embedding,which targets at learning the vector representation of vertices,has become a crucial issue in network analysis.However,considering the complex structures and heterogeneous attributes in real-world networks,existing methods may fail to handle the inconsistencies between the structure topology and attribute proximity.Thus,more comprehensive techniques are urgently required to capture the highly non-linear network structure and solve the existing inconsistencies with retaining more information.To that end,in this paper,we propose a heterogeneous-attributes enhancement deep framework(HEDF),which could better capture the non-linear structure and associated information in a deep learningway,and effectively combine the structure information of multi-views by the combining layer.Along this line,the inconsistencies will be handled to some extent and more structure information will be preserved through a semi-supervised mode.The extensive validations on several real-world datasets show that our model could outperform the baselines,especially for the sparse and inconsistent situation with less training data.
基金the National Key R&D Program of China(No.2021YFF0901003)。
文摘Recently advancements in deep learning models have significantly facilitated the development of sequential recommender systems(SRS).However,the current deep model structures are limited in their ability to learn high-quality embeddings with insufficient data.Meanwhile,highly skewed long-tail distribution is very common in recommender systems.Therefore,in this paper,we focus on enhancing the representation of tail items to improve sequential recommendation performance.Through empirical studies on benchmarks,we surprisingly observe that both the ranking performance and training procedure are greatly hindered by the poorly optimized tail item embeddings.To address this issue,we propose a sequential recommendation framework named TailRec that enables contextual information of tail item well-leveraged and greatly improves its corresponding representation.Given the characteristics of the sequential recommendation task,the surrounding interaction records of each tail item are regarded as contextual information without leveraging any additional side information.This approach allows for the mining of contextual information from cross-sequence behaviors to boost the performance of sequential recommendations.Such a light contextual filtering component is plug-and-play for a series of SRS models.To verify the effectiveness of the proposed TailRec,we conduct extensive experiments over several popular benchmark recommenders.The experimental results demonstrate that TailRec can greatly improve the recommendation results and speed up the training process.The codes of our methods have been available.
基金This work was supported by the National Natural Science Foundation of China(Grant No.62071378)the Shaanxi Province International Science and Technology Cooperation Program(2022KW-04)the Xi’an Science and Technology Plan Project(21XJZZ0072).
文摘Vehicle Color Recognition(VCR)plays a vital role in intelligent traffic management and criminal investigation assistance.However,the existing vehicle color datasets only cover 13 classes,which can not meet the current actual demand.Besides,although lots of efforts are devoted to VCR,they suffer from the problem of class imbalance in datasets.To address these challenges,in this paper,we propose a novel VCR method based on Smooth Modulation Neural Network with Multi-Scale Feature Fusion(SMNN-MSFF).Specifically,to construct the benchmark of model training and evaluation,we first present a new VCR dataset with 24 vehicle classes,Vehicle Color-24,consisting of 10091 vehicle images from a 100-hour urban road surveillance video.Then,to tackle the problem of long-tail distribution and improve the recognition performance,we propose the SMNN-MSFF model with multiscale feature fusion and smooth modulation.The former aims to extract feature information from local to global,and the latter could increase the loss of the images of tail class instances for training with class-imbalance.Finally,comprehensive experimental evaluation on Vehicle Color-24 and previously three representative datasets demonstrate that our proposed SMNN-MSFF outperformed state-of-the-art VCR methods.And extensive ablation studies also demonstrate that each module of our method is effective,especially,the smooth modulation efficiently help feature learning of the minority or tail classes.Vehicle Color-24 and the code of SMNN-MSFF are publicly available and can contact the author to obtain.
基金supported in part by the National Natural Science Foundation of China(Grant Nos.62406264 and 61727809)the project of University of Science and Technology of China-Huadong Photoelectric Co.,Ltd.,China(Grant No.OSHT-202311003)Anhui Provincial Science and Technology Major Project,China(Grant No.202203a05020042).
文摘With the increasing frequency of natural disasters and health emergencies,wearable infrared thermal imaging devices are becoming more prevalent in fire protection and medical fields.However,these devices often face imaging performance challenges such as insufficient contrast,dark areas and blurred edges,which significantly limit their practical effectiveness.To tackle these challenges,we propose a novel unsupervised lightweight 3D convolutional network(UL3DCN)specifically designed for enhancing infrared images on wearable devices.In this framework,the task of infrared image enhancement is conceptualized as generating high dynamic range infrared images from the corresponding temperature sequences during thermal equilibrium.To achieve this,we first design a learnable dynamic filtering module tailored for simulating a series of infrared image sequences under varying temperature differences.This module extends a single image from the spatial domain into the spatio-temporal domain.Subsequently,we employ a lightweight 3D convolution module to effectively extract spatio-temporal information from the image sequence.Finally,inspired by Zero-DCE,we utilize the extracted information to estimate pixel values and highorder curves,thereby enhancing the infrared images.Comprehensive experimental results demonstrate that our method achieves outstanding performance and real-time capabilities.Additionally,the proposed UL3DCN model has been successfully integrated into a wearable infrared firefighting mask.
基金funded by the National Natural Science Foundation of China(Nos.U23A20319,62441227,62441239,62472394,62202443,and 62506352)the Anhui Province Science and Technology Innovation Project(No.202423k09020011)the Anhui Provincial Science and Technology Major Project(No.2023z020006).
文摘Recently,with the rapid advancements in Large Language Models(LLMs),LLM-based Open-domain Question Answering(OpenQA)methods have reaped the benefits of emergent understanding and answering capabilities enabled by massive parameters compared to traditional methods.However,most of these methods encounter two critical challenges:how to integrate knowledge into LLMs effectively and how to adaptively generate results with specific answer formats.To address these challenges,we propose a novel framework,which aims to improve the OpenQA performance by exploring knowledge integration and controllable generation on LLMs simultaneously,namely GenKI.Specifically,we first train a dense passage retrieval model to retrieve associated knowledge from a given knowledge base.Subsequently,we introduce a novel knowledge integration model that incorporates the retrieval knowledge into instructions during fine-tuning to intensify the model.Furthermore,to enable controllable generation in LLMs,we leverage a certain fine-tuned LLM and an ensemble framework based on text consistency incorporating all coherence,fluency,and answer format assurance.Finally,extensive experiments conducted on three datasets with diverse answer formats demonstrate the effectiveness of GenKI with comparison of state-of-the-art baselines.Moreover,ablation studies have disclosed a linear relationship between the frequency of retrieved knowledge and the model’s ability to recall knowledge accurately with the ground truth.Tests focusing on the out-of-domain scenario and knowledge base independence scenario have further affirmed the robustness and controllable capability of GenKI.Our code of GenKI is available at https://github.com/USTC-StarTeam/GenKI.
