The distributed passive measurement is an important technology for networkbehavior research. To achieve a consistent measurement, the same packets should be sampled atdistributed measurement points. And in order to es...The distributed passive measurement is an important technology for networkbehavior research. To achieve a consistent measurement, the same packets should be sampled atdistributed measurement points. And in order to estimate the character of traffic statistics, thetraffic sample should be random in statistics. A distributed samplingmask measurement model isintroduced to tackle the difficulty of measuring the full trace of high-speed networks. The keypoint of the model is to choose some bits that are suitable to be sampling mask. In the paper, thebit entropy and bit flow entropy of IP packet headers in CERNET backbone are analyzed, and we findthat the 16 bits of identification field in IP packet header are fit to the matching field ofsampling mask. Measurement traffic also can be used to analyze the statistical character ofmeasurement sample and the randomicity of the model. At the same time the experiment resultsindicate that the model has a good sampling performance.展开更多
A novel dynamic batch selective sampling algorithm based on version space analysis is presented. In the traditional batch selective sampling, example selection is entirely determined by the existing unreliable classif...A novel dynamic batch selective sampling algorithm based on version space analysis is presented. In the traditional batch selective sampling, example selection is entirely determined by the existing unreliable classification boundary; meanwhile, within a batch, examples labeled previously fail to provide instructive information for the selection of the rest. As a result, using the examples selected in batch mode for model refinement will jeopardize the classification performance. Based on the duality between feature space and parameter space under the SVM active learning fi：amework, dynamic batch selective sampling is proposed to address the problem. We select a batch of examples dynamically, using the examples labeled previously as guidance for further selection. In this way, the selection of feedback examples is determined by both the existing classification model and the examples labeled previously. Encouraging experimental results demonstrate the effectiveness of the proposed algorithm.展开更多
Antarctica plays a key role in global energy balance and sea level change.It has been conventionally viewed as a whole ice body with high albedo in General Circulation Models or Regional Climate Models and the differe...Antarctica plays a key role in global energy balance and sea level change.It has been conventionally viewed as a whole ice body with high albedo in General Circulation Models or Regional Climate Models and the differences of land cover has usually been overlooked.Land cover in Antarctica is one of the most important drivers of changes in the Earth system.Detailed land cover information over the Antarctic region is necessary as spatial resolution improves in land process models.However,there is a lack of complete Antarctic land cover dataset derived from a consistent data source.To fill this data gap,we have produced a database named Antarctic Land Cover Database for the Year 2000(AntarcticaLC2000) using Landsat Enhanced Thematic Mapper Plus(ETM+) data acquired around 2000 and Moderate Resolution Imaging Spectrometer(MODIS) images acquired in the austral summer of 2003/2004 according to the criteria for the 1:100000-scale.Three land cover types were included in this map,separately,ice-free rocks,blue ice,and snow/firn.This classification legend was determined based on a review of the land cover systems in Antarctica(LCCSA) and an analysis of different land surface types and the potential of satellite data.Image classification was conducted through a combined usage of computer-aided and manual interpretation methods.A total of 4067 validation sample units were collected through visual interpretation in a stratified random sampling manner.An overall accuracy of 92.3%and the Kappa coefficient of 0.836 were achieved.Results show that the areas and percentages of ice-free rocks,blue ice,and snow/firn are 73268.81 km2(0.537%),225937.26 km2(1.656%),and 13345460.41 km2(97.807%),respectively.The comparisons with other different data proved a higher accuracy of our product and a more advantageous data quality.These indicate that AntarcticaLC2000,the new land cover dataset for Antarctica entirely derived from satellite data,is a reliable product for a broad spectrum of applications.展开更多
文摘The distributed passive measurement is an important technology for networkbehavior research. To achieve a consistent measurement, the same packets should be sampled atdistributed measurement points. And in order to estimate the character of traffic statistics, thetraffic sample should be random in statistics. A distributed samplingmask measurement model isintroduced to tackle the difficulty of measuring the full trace of high-speed networks. The keypoint of the model is to choose some bits that are suitable to be sampling mask. In the paper, thebit entropy and bit flow entropy of IP packet headers in CERNET backbone are analyzed, and we findthat the 16 bits of identification field in IP packet header are fit to the matching field ofsampling mask. Measurement traffic also can be used to analyze the statistical character ofmeasurement sample and the randomicity of the model. At the same time the experiment resultsindicate that the model has a good sampling performance.
文摘A novel dynamic batch selective sampling algorithm based on version space analysis is presented. In the traditional batch selective sampling, example selection is entirely determined by the existing unreliable classification boundary; meanwhile, within a batch, examples labeled previously fail to provide instructive information for the selection of the rest. As a result, using the examples selected in batch mode for model refinement will jeopardize the classification performance. Based on the duality between feature space and parameter space under the SVM active learning fi：amework, dynamic batch selective sampling is proposed to address the problem. We select a batch of examples dynamically, using the examples labeled previously as guidance for further selection. In this way, the selection of feedback examples is determined by both the existing classification model and the examples labeled previously. Encouraging experimental results demonstrate the effectiveness of the proposed algorithm.
基金supported by the Chinese Arctic and Antarctic Administration.National Basic Research Program of China(Grant No.2012CB957704)National Natural Science Foundation of China(Grant Nos.41676176 & 41676182)National High-tech R&D Program of China(Grant No.2008AA09Z117)
文摘Antarctica plays a key role in global energy balance and sea level change.It has been conventionally viewed as a whole ice body with high albedo in General Circulation Models or Regional Climate Models and the differences of land cover has usually been overlooked.Land cover in Antarctica is one of the most important drivers of changes in the Earth system.Detailed land cover information over the Antarctic region is necessary as spatial resolution improves in land process models.However,there is a lack of complete Antarctic land cover dataset derived from a consistent data source.To fill this data gap,we have produced a database named Antarctic Land Cover Database for the Year 2000(AntarcticaLC2000) using Landsat Enhanced Thematic Mapper Plus(ETM+) data acquired around 2000 and Moderate Resolution Imaging Spectrometer(MODIS) images acquired in the austral summer of 2003/2004 according to the criteria for the 1:100000-scale.Three land cover types were included in this map,separately,ice-free rocks,blue ice,and snow/firn.This classification legend was determined based on a review of the land cover systems in Antarctica(LCCSA) and an analysis of different land surface types and the potential of satellite data.Image classification was conducted through a combined usage of computer-aided and manual interpretation methods.A total of 4067 validation sample units were collected through visual interpretation in a stratified random sampling manner.An overall accuracy of 92.3%and the Kappa coefficient of 0.836 were achieved.Results show that the areas and percentages of ice-free rocks,blue ice,and snow/firn are 73268.81 km2(0.537%),225937.26 km2(1.656%),and 13345460.41 km2(97.807%),respectively.The comparisons with other different data proved a higher accuracy of our product and a more advantageous data quality.These indicate that AntarcticaLC2000,the new land cover dataset for Antarctica entirely derived from satellite data,is a reliable product for a broad spectrum of applications.
