Using the year-to-year increment approach,this study investigated the relationship of selected climatic elements with the increment time series of the summer rainfall between successive years in Northeast China,includ...Using the year-to-year increment approach,this study investigated the relationship of selected climatic elements with the increment time series of the summer rainfall between successive years in Northeast China,including the soil moisture content,sea surface temperature,500 hPa geopotential height,and sea level pressure in the preceding spring for the period 1981-2008.Two spring predictors were used to construct the seasonal prediction model:the area mean soil moisture content in Northwest Eurasia and the 500 hPa geopotential height over Northeast China.Both the cross-validation and comparison with previous studies showed that the above two predictors have good predicting ability for the summer rainfall in Northeast China.展开更多
Humans are experiencing the inclusion of artificial agents in their lives,such as unmanned vehicles,service robots,voice assistants,and intelligent medical care.If the artificial agents cannot align with social values...Humans are experiencing the inclusion of artificial agents in their lives,such as unmanned vehicles,service robots,voice assistants,and intelligent medical care.If the artificial agents cannot align with social values or make ethical decisions,they may not meet the expectations of humans.Traditionally,an ethical decision-making framework is constructed by rule-based or statistical approaches.In this paper,we propose an ethical decision-making framework based on incremental ILP(Inductive Logic Programming),which can overcome the brittleness of rule-based approaches and little interpretability of statistical approaches.As the current incremental ILP makes it difficult to solve conflicts,we propose a novel ethical decision-making framework considering conflicts in this paper,which adopts our proposed incremental ILP system.The framework consists of two processes:the learning process and the deduction process.The first process records bottom clauses with their score functions and learns rules guided by the entailment and the score function.The second process obtains an ethical decision based on the rules.In an ethical scenario about chatbots for teenagers’mental health,we verify that our framework can learn ethical rules and make ethical decisions.Besides,we extract incremental ILP from the framework and compare it with the state-of-the-art ILP systems based on ASP(Answer Set Programming)focusing on conflict resolution.The results of comparisons show that our proposed system can generate better-quality rules than most other systems.展开更多
Multispecies forests have received increased scientific attention,driven by the hypothesis that biodiversity improves ecological resilience.However,a greater species diversity presents challenges for forest management...Multispecies forests have received increased scientific attention,driven by the hypothesis that biodiversity improves ecological resilience.However,a greater species diversity presents challenges for forest management and research.Our study aims to develop basal area growth models for tree species cohorts.The analysis is based on a dataset of 423 permanent plots(2,500 m^(2))located in temperate forests in Durango,Mexico.First,we define tree species cohorts based on individual and neighborhood-based variables using a combination of principal component and cluster analyses.Then,we estimate the basal area increment of each cohort through the generalized additive model to describe the effect of tree size,competition,stand density and site quality.The principal component and cluster analyses assign a total of 37 tree species to eight cohorts that differed primarily with regard to the distribution of tree size and vertical position within the community.The generalized additive models provide satisfactory estimates of tree growth for the species cohorts,explaining between 19 and 53 percent of the total variation of basal area increment,and highlight the following results:i)most cohorts show a"rise-and-fall"effect of tree size on tree growth;ii)surprisingly,the competition index"basal area of larger trees"had showed a positive effect in four of the eight cohorts;iii)stand density had a negative effect on basal area increment,though the effect was minor in medium-and high-density stands,and iv)basal area growth was positively correlated with site quality except for an oak cohort.The developed species cohorts and growth models provide insight into their particular ecological features and growth patterns that may support the development of sustainable management strategies for temperate multispecies forests.展开更多
The visions of Industry 4.0 and 5.0 have reinforced the industrial environment.They have also made artificial intelligence incorporated as a major facilitator.Diagnosing machine faults has become a solid foundation fo...The visions of Industry 4.0 and 5.0 have reinforced the industrial environment.They have also made artificial intelligence incorporated as a major facilitator.Diagnosing machine faults has become a solid foundation for automatically recognizing machine failure,and thus timely maintenance can ensure safe operations.Transfer learning is a promising solution that can enhance the machine fault diagnosis model by borrowing pre-trained knowledge from the source model and applying it to the target model,which typically involves two datasets.In response to the availability of multiple datasets,this paper proposes using selective and adaptive incremental transfer learning(SA-ITL),which fuses three algorithms,namely,the hybrid selective algorithm,the transferability enhancement algorithm,and the incremental transfer learning algorithm.It is a selective algorithm that enables selecting and ordering appropriate datasets for transfer learning and selecting useful knowledge to avoid negative transfer.The algorithm also adaptively adjusts the portion of training data to balance the learning rate and training time.The proposed algorithm is evaluated and analyzed using ten benchmark datasets.Compared with other algorithms from existing works,SA-ITL improves the accuracy of all datasets.Ablation studies present the accuracy enhancements of the SA-ITL,including the hybrid selective algorithm(1.22%-3.82%),transferability enhancement algorithm(1.91%-4.15%),and incremental transfer learning algorithm(0.605%-2.68%).These also show the benefits of enhancing the target model with heterogeneous image datasets that widen the range of domain selection between source and target domains.展开更多
We investigated the parametric optimization on incremental sheet forming of stainless steel using Grey Relational Analysis(GRA) coupled with Principal Component Analysis(PCA). AISI 316L stainless steel sheets were use...We investigated the parametric optimization on incremental sheet forming of stainless steel using Grey Relational Analysis(GRA) coupled with Principal Component Analysis(PCA). AISI 316L stainless steel sheets were used to develop double wall angle pyramid with aid of tungsten carbide tool. GRA coupled with PCA was used to plan the experiment conditions. Control factors such as Tool Diameter(TD), Step Depth(SD), Bottom Wall Angle(BWA), Feed Rate(FR) and Spindle Speed(SS) on Top Wall Angle(TWA) and Top Wall Angle Surface Roughness(TWASR) have been studied. Wall angle increases with increasing tool diameter due to large contact area between tool and workpiece. As the step depth, feed rate and spindle speed increase,TWASR decreases with increasing tool diameter. As the step depth increasing, the hydrostatic stress is raised causing severe cracks in the deformed surface. Hence it was concluded that the proposed hybrid method was suitable for optimizing the factors and response.展开更多
We present a model for predicting summertime surface air temperature in Northeast China(NESSAT) using a year-to-year incremental approach.The predicted value for each year's increase or decrease of NESSAT is added ...We present a model for predicting summertime surface air temperature in Northeast China(NESSAT) using a year-to-year incremental approach.The predicted value for each year's increase or decrease of NESSAT is added to the observed value within a particular year to yield the net forecast NESSAT.The seasonal forecast model for the year-to-year increments of NESSAT is constructed based on data from 1975- 2007.Five predictors are used:an index for sea ice cover over the East Siberian Sea,an index for central Pacific tropical sea surface temperature,two high latitude circulation indices,as well as a North American pressure index.All predictors are available by no later than March,which allows for compilation of a seasonal forecast with a two-month lead time.The prediction model accurately captures the interannual variations of NESSAT during 1977-2007 with a correlation coefficient between the predicted and observed NESSAT of 0.87(accounting for 76%of total variance) and a mean absolute error(MAE) of 0.3℃.A cross-validation test during 1977-2008 demonstrates that the model has good predictive skill,with MAE of 0.4℃and a correlation coefficient between the predicted and observed NESSAT of 0.76.展开更多
A new approach to forecasting the year-to-year increment of rainfall in North China in July-August (JA) is proposed. DY is defined as the difference of a variable between the current year and the preceding year (year-...A new approach to forecasting the year-to-year increment of rainfall in North China in July-August (JA) is proposed. DY is defined as the difference of a variable between the current year and the preceding year (year-to-year increment). NR denotes the seasonal mean precipitation rate over North China in JA. After analyzing the atmospheric circulation anomalies associated with the DY of NR, five key predictors for the DY of NR have been identified. The prediction model for the DY of NR is established by using multi-linear regression method and the NR is obtained (the current forecasted DY of NR added to the preceding observed NR). The prediction model shows a high correlation coefficient (0.8) between the simulated and the observed DY of NR throughout period 1965-1999, with an average relative root mean square error of 19% for the percentage of precipitation rate anomaly over North China. The prediction model makes a hindcast for 2000-2007, with an average relative root mean square error of 21% for the percentage of precipitation rate anomaly over North China. The model reproduces the downward trend of the percentage of precipitation rate anomaly over North China during 1965-2006. Because the current operational prediction models of the summer precipitation have average forecast scores of 60%-70%, it has been more difficult to forecast the summer rainfall over North China. Thus this new approach for predicting the year-to-year increment of the summer precipitation (and hence the summer precipitation itself) has the potential to significantly improve operational forecasting skill for summer precipitation.展开更多
Deep Convolution Neural Networks(DCNNs)can capture discriminative features from large datasets.However,how to incrementally learn new samples without forgetting old ones and recognize novel classes that arise in the d...Deep Convolution Neural Networks(DCNNs)can capture discriminative features from large datasets.However,how to incrementally learn new samples without forgetting old ones and recognize novel classes that arise in the dynamically changing world,e.g.,classifying newly discovered fish species,remains an open problem.We address an even more challenging and realistic setting of this problem where new class samples are insufficient,i.e.,Few-Shot Class-Incremental Learning(FSCIL).Current FSCIL methods augment the training data to alleviate the overfitting of novel classes.By contrast,we propose Filter Bank Networks(FBNs)that augment the learnable filters to capture fine-detailed features for adapting to future new classes.In the forward pass,FBNs augment each convolutional filter to a virtual filter bank containing the canonical one,i.e.,itself,and multiple transformed versions.During back-propagation,FBNs explicitly stimulate fine-detailed features to emerge and collectively align all gradients of each filter bank to learn the canonical one.FBNs capture pattern variants that do not yet exist in the pretraining session,thus making it easy to incorporate new classes in the incremental learning phase.Moreover,FBNs introduce model-level prior knowledge to efficiently utilize the limited few-shot data.Extensive experiments on MNIST,CIFAR100,CUB200,andMini-ImageNet datasets show that FBNs consistently outperformthe baseline by a significantmargin,reporting new state-of-the-art FSCIL results.In addition,we contribute a challenging FSCIL benchmark,Fishshot1K,which contains 8261 underwater images covering 1000 ocean fish species.The code is included in the supplementary materials.展开更多
Recently, deep convolutional neural networks (DCNNs) have achieved remarkable results in image classification tasks. Despite convolutional networks’ great successes, their training process relies on a large amount of...Recently, deep convolutional neural networks (DCNNs) have achieved remarkable results in image classification tasks. Despite convolutional networks’ great successes, their training process relies on a large amount of data prepared in advance, which is often challenging in real-world applications, such as streaming data and concept drift. For this reason, incremental learning (continual learning) has attracted increasing attention from scholars. However, incremental learning is associated with the challenge of catastrophic forgetting: the performance on previous tasks drastically degrades after learning a new task. In this paper, we propose a new strategy to alleviate catastrophic forgetting when neural networks are trained in continual domains. Specifically, two components are applied: data translation based on transfer learning and knowledge distillation. The former translates a portion of new data to reconstruct the partial data distribution of the old domain. The latter uses an old model as a teacher to guide a new model. The experimental results on three datasets have shown that our work can effectively alleviate catastrophic forgetting by a combination of the two methods aforementioned.展开更多
Background With the development of information technology,there is a significant increase in the number of network traffic logs mixed with various types of cyberattacks.Traditional intrusion detection systems(IDSs)are...Background With the development of information technology,there is a significant increase in the number of network traffic logs mixed with various types of cyberattacks.Traditional intrusion detection systems(IDSs)are limited in detecting new inconstant patterns and identifying malicious traffic traces in real time.Therefore,there is an urgent need to implement more effective intrusion detection technologies to protect computer security.Methods In this study,we designed a hybrid IDS by combining our incremental learning model(KANSOINN)and active learning to learn new log patterns and detect various network anomalies in real time.Conclusions Experimental results on the NSLKDD dataset showed that KAN-SOINN can be continuously improved and effectively detect malicious logs.Meanwhile,comparative experiments proved that using a hybrid query strategy in active learning can improve the model learning efficiency.展开更多
Attribute reduction,also known as feature selection,for decision information systems is one of the most pivotal issues in machine learning and data mining.Approaches based on the rough set theory and some extensions w...Attribute reduction,also known as feature selection,for decision information systems is one of the most pivotal issues in machine learning and data mining.Approaches based on the rough set theory and some extensions were proved to be efficient for dealing with the problemof attribute reduction.Unfortunately,the intuitionistic fuzzy sets based methods have not received much interest,while these methods are well-known as a very powerful approach to noisy decision tables,i.e.,data tables with the low initial classification accuracy.Therefore,this paper provides a novel incremental attribute reductionmethod to dealmore effectivelywith noisy decision tables,especially for highdimensional ones.In particular,we define a new reduct and then design an original attribute reduction method based on the distance measure between two intuitionistic fuzzy partitions.It should be noted that the intuitionistic fuzzypartitiondistance iswell-knownas aneffectivemeasure todetermine important attributes.More interestingly,an incremental formula is also developed to quickly compute the intuitionistic fuzzy partition distance in case when the decision table increases in the number of objects.This formula is then applied to construct an incremental attribute reduction algorithm for handling such dynamic tables.Besides,some experiments are conducted on real datasets to show that our method is far superior to the fuzzy rough set based methods in terms of the size of reduct and the classification accuracy.展开更多
In the traditional incremental analysis update(IAU)process,all analysis increments are treated as constant forcing in a model’s prognostic equations over a certain time window.This approach effectively reduces high-f...In the traditional incremental analysis update(IAU)process,all analysis increments are treated as constant forcing in a model’s prognostic equations over a certain time window.This approach effectively reduces high-frequency oscillations introduced by data assimilation.However,as different scales of increments have unique evolutionary speeds and life histories in a numerical model,the traditional IAU scheme cannot fully meet the requirements of short-term forecasting for the damping of high-frequency noise and may even cause systematic drifts.Therefore,a multi-scale IAU scheme is proposed in this paper.Analysis increments were divided into different scale parts using a spatial filtering technique.For each scale increment,the optimal relaxation time in the IAU scheme was determined by the skill of the forecasting results.Finally,different scales of analysis increments were added to the model integration during their optimal relaxation time.The multi-scale IAU scheme can effectively reduce the noise and further improve the balance between large-scale and small-scale increments in the model initialization stage.To evaluate its performance,several numerical experiments were conducted to simulate the path and intensity of Typhoon Mangkhut(2018)and showed that:(1)the multi-scale IAU scheme had an obvious effect on noise control at the initial stage of data assimilation;(2)the optimal relaxation time for large-scale and small-scale increments was estimated as 6 h and 3 h,respectively;(3)the forecast performance of the multi-scale IAU scheme in the prediction of Typhoon Mangkhut(2018)was better than that of the traditional IAU scheme.The results demonstrate the superiority of the multi-scale IAU scheme.展开更多
基金supported by the National Basic Research Program of China under Grants 2010CB950304 and 2009CB421406the Special Fund for the public welfare indus-try (Meteorology) under Grant GYHY200906018+1 种基金the Knowledge Innovation Program of the Chinese Academy of Sciences under Grant KZCX2-YW-QN202the Chinese Academy of Sciences under Grants KZCX2-YW-Q1-02 and KZCX2-YW-Q11-00
文摘Using the year-to-year increment approach,this study investigated the relationship of selected climatic elements with the increment time series of the summer rainfall between successive years in Northeast China,including the soil moisture content,sea surface temperature,500 hPa geopotential height,and sea level pressure in the preceding spring for the period 1981-2008.Two spring predictors were used to construct the seasonal prediction model:the area mean soil moisture content in Northwest Eurasia and the 500 hPa geopotential height over Northeast China.Both the cross-validation and comparison with previous studies showed that the above two predictors have good predicting ability for the summer rainfall in Northeast China.
基金This work was funded by the National Natural Science Foundation of China Nos.U22A2099,61966009,62006057the Graduate Innovation Program No.YCSW2022286.
文摘Humans are experiencing the inclusion of artificial agents in their lives,such as unmanned vehicles,service robots,voice assistants,and intelligent medical care.If the artificial agents cannot align with social values or make ethical decisions,they may not meet the expectations of humans.Traditionally,an ethical decision-making framework is constructed by rule-based or statistical approaches.In this paper,we propose an ethical decision-making framework based on incremental ILP(Inductive Logic Programming),which can overcome the brittleness of rule-based approaches and little interpretability of statistical approaches.As the current incremental ILP makes it difficult to solve conflicts,we propose a novel ethical decision-making framework considering conflicts in this paper,which adopts our proposed incremental ILP system.The framework consists of two processes:the learning process and the deduction process.The first process records bottom clauses with their score functions and learns rules guided by the entailment and the score function.The second process obtains an ethical decision based on the rules.In an ethical scenario about chatbots for teenagers’mental health,we verify that our framework can learn ethical rules and make ethical decisions.Besides,we extract incremental ILP from the framework and compare it with the state-of-the-art ILP systems based on ASP(Answer Set Programming)focusing on conflict resolution.The results of comparisons show that our proposed system can generate better-quality rules than most other systems.
基金The National Forestry Commission of Mexico and The Mexican National Council for Science and Technology(CONAFOR-CONACYT-115900)。
文摘Multispecies forests have received increased scientific attention,driven by the hypothesis that biodiversity improves ecological resilience.However,a greater species diversity presents challenges for forest management and research.Our study aims to develop basal area growth models for tree species cohorts.The analysis is based on a dataset of 423 permanent plots(2,500 m^(2))located in temperate forests in Durango,Mexico.First,we define tree species cohorts based on individual and neighborhood-based variables using a combination of principal component and cluster analyses.Then,we estimate the basal area increment of each cohort through the generalized additive model to describe the effect of tree size,competition,stand density and site quality.The principal component and cluster analyses assign a total of 37 tree species to eight cohorts that differed primarily with regard to the distribution of tree size and vertical position within the community.The generalized additive models provide satisfactory estimates of tree growth for the species cohorts,explaining between 19 and 53 percent of the total variation of basal area increment,and highlight the following results:i)most cohorts show a"rise-and-fall"effect of tree size on tree growth;ii)surprisingly,the competition index"basal area of larger trees"had showed a positive effect in four of the eight cohorts;iii)stand density had a negative effect on basal area increment,though the effect was minor in medium-and high-density stands,and iv)basal area growth was positively correlated with site quality except for an oak cohort.The developed species cohorts and growth models provide insight into their particular ecological features and growth patterns that may support the development of sustainable management strategies for temperate multispecies forests.
文摘The visions of Industry 4.0 and 5.0 have reinforced the industrial environment.They have also made artificial intelligence incorporated as a major facilitator.Diagnosing machine faults has become a solid foundation for automatically recognizing machine failure,and thus timely maintenance can ensure safe operations.Transfer learning is a promising solution that can enhance the machine fault diagnosis model by borrowing pre-trained knowledge from the source model and applying it to the target model,which typically involves two datasets.In response to the availability of multiple datasets,this paper proposes using selective and adaptive incremental transfer learning(SA-ITL),which fuses three algorithms,namely,the hybrid selective algorithm,the transferability enhancement algorithm,and the incremental transfer learning algorithm.It is a selective algorithm that enables selecting and ordering appropriate datasets for transfer learning and selecting useful knowledge to avoid negative transfer.The algorithm also adaptively adjusts the portion of training data to balance the learning rate and training time.The proposed algorithm is evaluated and analyzed using ten benchmark datasets.Compared with other algorithms from existing works,SA-ITL improves the accuracy of all datasets.Ablation studies present the accuracy enhancements of the SA-ITL,including the hybrid selective algorithm(1.22%-3.82%),transferability enhancement algorithm(1.91%-4.15%),and incremental transfer learning algorithm(0.605%-2.68%).These also show the benefits of enhancing the target model with heterogeneous image datasets that widen the range of domain selection between source and target domains.
文摘We investigated the parametric optimization on incremental sheet forming of stainless steel using Grey Relational Analysis(GRA) coupled with Principal Component Analysis(PCA). AISI 316L stainless steel sheets were used to develop double wall angle pyramid with aid of tungsten carbide tool. GRA coupled with PCA was used to plan the experiment conditions. Control factors such as Tool Diameter(TD), Step Depth(SD), Bottom Wall Angle(BWA), Feed Rate(FR) and Spindle Speed(SS) on Top Wall Angle(TWA) and Top Wall Angle Surface Roughness(TWASR) have been studied. Wall angle increases with increasing tool diameter due to large contact area between tool and workpiece. As the step depth, feed rate and spindle speed increase,TWASR decreases with increasing tool diameter. As the step depth increasing, the hydrostatic stress is raised causing severe cracks in the deformed surface. Hence it was concluded that the proposed hybrid method was suitable for optimizing the factors and response.
基金Supported by the Special Fund for Public Welfare(Meteorology)(GYHY200906018)the Innovation Key Program of the Chinese Academy of Sciences(KZCX2-YW-BR-14)+1 种基金the Basic Research Program of China(2009CB421406)the National Excellent Ph.D.Dissertation Program of the Chinese Academy of Sciences
文摘We present a model for predicting summertime surface air temperature in Northeast China(NESSAT) using a year-to-year incremental approach.The predicted value for each year's increase or decrease of NESSAT is added to the observed value within a particular year to yield the net forecast NESSAT.The seasonal forecast model for the year-to-year increments of NESSAT is constructed based on data from 1975- 2007.Five predictors are used:an index for sea ice cover over the East Siberian Sea,an index for central Pacific tropical sea surface temperature,two high latitude circulation indices,as well as a North American pressure index.All predictors are available by no later than March,which allows for compilation of a seasonal forecast with a two-month lead time.The prediction model accurately captures the interannual variations of NESSAT during 1977-2007 with a correlation coefficient between the predicted and observed NESSAT of 0.87(accounting for 76%of total variance) and a mean absolute error(MAE) of 0.3℃.A cross-validation test during 1977-2008 demonstrates that the model has good predictive skill,with MAE of 0.4℃and a correlation coefficient between the predicted and observed NESSAT of 0.76.
基金Supported by National Basic Research Program of China (Grant No. 2009CB421406)National Natural Science Foundation of China (Grant Nos. 40631005, 40775049)Excellent Ph. D Dissertation in Chinese Academy of Sciences
文摘A new approach to forecasting the year-to-year increment of rainfall in North China in July-August (JA) is proposed. DY is defined as the difference of a variable between the current year and the preceding year (year-to-year increment). NR denotes the seasonal mean precipitation rate over North China in JA. After analyzing the atmospheric circulation anomalies associated with the DY of NR, five key predictors for the DY of NR have been identified. The prediction model for the DY of NR is established by using multi-linear regression method and the NR is obtained (the current forecasted DY of NR added to the preceding observed NR). The prediction model shows a high correlation coefficient (0.8) between the simulated and the observed DY of NR throughout period 1965-1999, with an average relative root mean square error of 19% for the percentage of precipitation rate anomaly over North China. The prediction model makes a hindcast for 2000-2007, with an average relative root mean square error of 21% for the percentage of precipitation rate anomaly over North China. The model reproduces the downward trend of the percentage of precipitation rate anomaly over North China during 1965-2006. Because the current operational prediction models of the summer precipitation have average forecast scores of 60%-70%, it has been more difficult to forecast the summer rainfall over North China. Thus this new approach for predicting the year-to-year increment of the summer precipitation (and hence the summer precipitation itself) has the potential to significantly improve operational forecasting skill for summer precipitation.
基金support from the Strategic Priority Research Program of the Chinese Academy of Sciences under Grant No.XDA27000000.
文摘Deep Convolution Neural Networks(DCNNs)can capture discriminative features from large datasets.However,how to incrementally learn new samples without forgetting old ones and recognize novel classes that arise in the dynamically changing world,e.g.,classifying newly discovered fish species,remains an open problem.We address an even more challenging and realistic setting of this problem where new class samples are insufficient,i.e.,Few-Shot Class-Incremental Learning(FSCIL).Current FSCIL methods augment the training data to alleviate the overfitting of novel classes.By contrast,we propose Filter Bank Networks(FBNs)that augment the learnable filters to capture fine-detailed features for adapting to future new classes.In the forward pass,FBNs augment each convolutional filter to a virtual filter bank containing the canonical one,i.e.,itself,and multiple transformed versions.During back-propagation,FBNs explicitly stimulate fine-detailed features to emerge and collectively align all gradients of each filter bank to learn the canonical one.FBNs capture pattern variants that do not yet exist in the pretraining session,thus making it easy to incorporate new classes in the incremental learning phase.Moreover,FBNs introduce model-level prior knowledge to efficiently utilize the limited few-shot data.Extensive experiments on MNIST,CIFAR100,CUB200,andMini-ImageNet datasets show that FBNs consistently outperformthe baseline by a significantmargin,reporting new state-of-the-art FSCIL results.In addition,we contribute a challenging FSCIL benchmark,Fishshot1K,which contains 8261 underwater images covering 1000 ocean fish species.The code is included in the supplementary materials.
文摘Recently, deep convolutional neural networks (DCNNs) have achieved remarkable results in image classification tasks. Despite convolutional networks’ great successes, their training process relies on a large amount of data prepared in advance, which is often challenging in real-world applications, such as streaming data and concept drift. For this reason, incremental learning (continual learning) has attracted increasing attention from scholars. However, incremental learning is associated with the challenge of catastrophic forgetting: the performance on previous tasks drastically degrades after learning a new task. In this paper, we propose a new strategy to alleviate catastrophic forgetting when neural networks are trained in continual domains. Specifically, two components are applied: data translation based on transfer learning and knowledge distillation. The former translates a portion of new data to reconstruct the partial data distribution of the old domain. The latter uses an old model as a teacher to guide a new model. The experimental results on three datasets have shown that our work can effectively alleviate catastrophic forgetting by a combination of the two methods aforementioned.
基金Supported by SJTU-HUAWEI TECH Cybersecurity Innovation Lab。
文摘Background With the development of information technology,there is a significant increase in the number of network traffic logs mixed with various types of cyberattacks.Traditional intrusion detection systems(IDSs)are limited in detecting new inconstant patterns and identifying malicious traffic traces in real time.Therefore,there is an urgent need to implement more effective intrusion detection technologies to protect computer security.Methods In this study,we designed a hybrid IDS by combining our incremental learning model(KANSOINN)and active learning to learn new log patterns and detect various network anomalies in real time.Conclusions Experimental results on the NSLKDD dataset showed that KAN-SOINN can be continuously improved and effectively detect malicious logs.Meanwhile,comparative experiments proved that using a hybrid query strategy in active learning can improve the model learning efficiency.
基金funded by Hanoi University of Industry under Grant Number 27-2022-RD/HD-DHCN (URL:https://www.haui.edu.vn/).
文摘Attribute reduction,also known as feature selection,for decision information systems is one of the most pivotal issues in machine learning and data mining.Approaches based on the rough set theory and some extensions were proved to be efficient for dealing with the problemof attribute reduction.Unfortunately,the intuitionistic fuzzy sets based methods have not received much interest,while these methods are well-known as a very powerful approach to noisy decision tables,i.e.,data tables with the low initial classification accuracy.Therefore,this paper provides a novel incremental attribute reductionmethod to dealmore effectivelywith noisy decision tables,especially for highdimensional ones.In particular,we define a new reduct and then design an original attribute reduction method based on the distance measure between two intuitionistic fuzzy partitions.It should be noted that the intuitionistic fuzzypartitiondistance iswell-knownas aneffectivemeasure todetermine important attributes.More interestingly,an incremental formula is also developed to quickly compute the intuitionistic fuzzy partition distance in case when the decision table increases in the number of objects.This formula is then applied to construct an incremental attribute reduction algorithm for handling such dynamic tables.Besides,some experiments are conducted on real datasets to show that our method is far superior to the fuzzy rough set based methods in terms of the size of reduct and the classification accuracy.
基金jointly sponsored by the Shenzhen Science and Technology Innovation Commission (Grant No. KCXFZ20201221173610028)the key program of the National Natural Science Foundation of China (Grant No. 42130605)
文摘In the traditional incremental analysis update(IAU)process,all analysis increments are treated as constant forcing in a model’s prognostic equations over a certain time window.This approach effectively reduces high-frequency oscillations introduced by data assimilation.However,as different scales of increments have unique evolutionary speeds and life histories in a numerical model,the traditional IAU scheme cannot fully meet the requirements of short-term forecasting for the damping of high-frequency noise and may even cause systematic drifts.Therefore,a multi-scale IAU scheme is proposed in this paper.Analysis increments were divided into different scale parts using a spatial filtering technique.For each scale increment,the optimal relaxation time in the IAU scheme was determined by the skill of the forecasting results.Finally,different scales of analysis increments were added to the model integration during their optimal relaxation time.The multi-scale IAU scheme can effectively reduce the noise and further improve the balance between large-scale and small-scale increments in the model initialization stage.To evaluate its performance,several numerical experiments were conducted to simulate the path and intensity of Typhoon Mangkhut(2018)and showed that:(1)the multi-scale IAU scheme had an obvious effect on noise control at the initial stage of data assimilation;(2)the optimal relaxation time for large-scale and small-scale increments was estimated as 6 h and 3 h,respectively;(3)the forecast performance of the multi-scale IAU scheme in the prediction of Typhoon Mangkhut(2018)was better than that of the traditional IAU scheme.The results demonstrate the superiority of the multi-scale IAU scheme.