Clustering is one of the recently challenging tasks since there is an ever.growing amount of data in scientific research and commercial applications. High quality and fast document clustering algorithms are in great d...Clustering is one of the recently challenging tasks since there is an ever.growing amount of data in scientific research and commercial applications. High quality and fast document clustering algorithms are in great demand to deal with large volume of data. The computational requirements for bringing such growing amount data to a central site for clustering are complex. The proposed algorithm uses optimal centroids for K.Means clustering based on Particle Swarm Optimization(PSO).PSO is used to take advantage of its global search ability to provide optimal centroids which aids in generating more compact clusters with improved accuracy. This proposed methodology utilizes Hadoop and Map Reduce framework which provides distributed storage and analysis to support data intensive distributed applications. Experiments were performed on Reuter's and RCV1 document dataset which shows an improvement in accuracy with reduced execution time.展开更多
In this paper,we investigate the spectrum sensing performance of a distributed satellite clusters(DSC)under perturbation,aiming to enhance the sensing ability of weak signals in the coexistence of strong and weak sign...In this paper,we investigate the spectrum sensing performance of a distributed satellite clusters(DSC)under perturbation,aiming to enhance the sensing ability of weak signals in the coexistence of strong and weak signals.Specifically,we propose a cooperative beamforming(BF)algorithm though random antenna array theory to fit the location characteristic of DSC and derive the average far-field beam pattern under perturbation.Then,a constrained optimization problem with maximizing the signal to interference plus noise ratio(SINR)is modeled to obtain the BF weight vectors,and an approximate expression of SINR is presented in the presence of the mismatch of signal steering vector.Finally,we derive the closedform expression of the detection probability for the considered DSC over Shadowed-Rician fading channels.Simulation results are provided to validate our theoretical analysis and to characterize the impact of various parameters on the system performance.展开更多
The rapid growth of modern mobile devices leads to a large number of distributed data,which is extremely valuable for learning models.Unfortunately,model training by collecting all these original data to a centralized...The rapid growth of modern mobile devices leads to a large number of distributed data,which is extremely valuable for learning models.Unfortunately,model training by collecting all these original data to a centralized cloud server is not applicable due to data privacy and communication costs concerns,hindering artificial intelligence from empowering mobile devices.Moreover,these data are not identically and independently distributed(Non-IID)caused by their different context,which will deteriorate the performance of the model.To address these issues,we propose a novel Distributed Learning algorithm based on hierarchical clustering and Adaptive Dataset Condensation,named ADC-DL,which learns a shared model by collecting the synthetic samples generated on each device.To tackle the heterogeneity of data distribution,we propose an entropy topsis comprehensive tiering model for hierarchical clustering,which distinguishes clients in terms of their data characteristics.Subsequently,synthetic dummy samples are generated based on the hierarchical structure utilizing adaptive dataset condensation.The procedure of dataset condensation can be adjusted adaptively according to the tier of the client.Extensive experiments demonstrate that the performance of our ADC-DL is more outstanding in prediction accuracy and communication costs compared with existing algorithms.展开更多
The reconstruction of spacecraft cluster based on local information and distributed strategy is investigated.Each spacecraft is an intelligent individual that can detect information within a limited range and can dete...The reconstruction of spacecraft cluster based on local information and distributed strategy is investigated.Each spacecraft is an intelligent individual that can detect information within a limited range and can determine its behavior based on surrounding information.The objective of the cluster is to achieve the formation reconstruction with minimum fuel consumption.Based on the principle of dual pulse rendezvous maneuver,three target selection strategies are designed for collision avoidance.Strategy-1 determines the target point’s attribution according to the target’s distance when the target point conflicts and uses a unit pulse to avoid a collision.Strategy-2 changes the collision avoidance behavior.When two spacecraft meet more than once,the strategy switches the target points of the two spacecraft.In Strategy-3,the spacecraft closer to the target has higher priority in target allocation.Strategy-3 also switches the target points when two spacecraft encounter more than once.The three strategies for a given position,different completion times,and random position are compared.Numerical simulations show that all three strategies can accomplish the spacecraft cluster's reconfiguration under the specified requirements.Strategy-3 is better than Strategy-1 in all simulation cases in the sense of less fuel consumption with different completion times and given location,and it is more effective than Strategy-2 in most of the completion time.With a random initial position and given time,Strategy-3 is better than Strategy-1 in about 70%of the cases and more stable.展开更多
In many applications such as computational fluid dynamics and weather prediction, as well as image processing and state of Markov chain etc., the grade of matrix n is often very large, and any serial algorithm cannot ...In many applications such as computational fluid dynamics and weather prediction, as well as image processing and state of Markov chain etc., the grade of matrix n is often very large, and any serial algorithm cannot solve the problems. A distributed cluster-based solution for very large linear equations is discussed, it includes the definitions of notations, partition of matrix, communication mechanism, and a master-slaver algorithm etc., the computing cost is O（n^3/N）, the memory cost is O（n^2/N）, the I/O cost is O（n^2/N）, and the com- munication cost is O（Nn ）, here, N is the number of computing nodes or processes. Some tests show that the solution could solve the double type of matrix under 10^6 × 10^6 effectively.展开更多
The sequential dispersing results of aerial cluster bomb are discussed. The ballistic model and the mod- el for impact point distribution of bullets are established. The main factors influencing impact point distribut...The sequential dispersing results of aerial cluster bomb are discussed. The ballistic model and the mod- el for impact point distribution of bullets are established. The main factors influencing impact point distribution are analyzed by numerical simulation. And the feasibility of improving distribution effect through sequential dis- persing is validated. Sequential dispersion and optimized airdrop parameters can help to achieve better battle effec- tiveness.展开更多
Wireless sensor network nodes （WSN nodes） have limited computing power, storage ca-pacity, conmmunication capabilities and energy and WSN nodes are easy to be paralyzed by Sybil at- tack. In order to prevent Sybil a...Wireless sensor network nodes （WSN nodes） have limited computing power, storage ca-pacity, conmmunication capabilities and energy and WSN nodes are easy to be paralyzed by Sybil at- tack. In order to prevent Sybil attacks, a new key distribution scheme for wireless sensor networks is presented. In this scheme, the key inforrmtion and node ID are associated, and then the attacker is dif-ficult to forge identity ID and the key inforrmtion corresponding to ID can not be forged. This scheme can use low-power to resist the Syhil attack and give full play to the resource advantages of the cluster head. The computing, storage and corrn^ni- cation is rminly undertaken by the cluster head o- verhead to achieve the lowest energy consumption and resist against nodes capture attack. Theoretical analysis and experimental results show that com- pared with the traditional scheme presented in Ref. [14], the capture rate of general nodes of cluster re-duces 40%, and the capture rate of cluster heads reduces 50%. So the scheme presented in this pa-per can improve resilience against nodes capture at- tack and reduce node power consumption.展开更多
For a collective system,the connectedness of the adjacency matrix plays a key role in making the system achieve its emergent feature,such as flocking or multi-clustering.In this paper,we study a nonsymmetric multi-par...For a collective system,the connectedness of the adjacency matrix plays a key role in making the system achieve its emergent feature,such as flocking or multi-clustering.In this paper,we study a nonsymmetric multi-particle system with a constant and local cut-off weight.A distributed communication delay is also introduced into both the velocity adjoint term and the cut-off weight.As a new observation,we show that the desired multi-particle system undergoes both flocking and clustering behaviors when the eigenvalue 1 of the adjacency matrix is semi-simple.In this case,the adjacency matrix may lose the connectedness.In particular,the number of clusters is discussed by using subspace analysis.In terms of results,for both the non-critical and general neighbourhood situations,some criteria of flocking and clustering emergence with an exponential convergent rate are established by the standard matrix analysis for when the delay is free.As a distributed delay is involved,the corresponding criteria are also found,and these small time lags do not change the emergent properties qualitatively,but alter the final value in a nonlinear way.Consequently,some previous works[14]are extended.展开更多
As a kind of statistical method, the technique of Hidden Markov Model （HMM） is widely used for speech recognition. In order to train the HMM to be more effective with much less amount of data, the Subspace Distribut...As a kind of statistical method, the technique of Hidden Markov Model （HMM） is widely used for speech recognition. In order to train the HMM to be more effective with much less amount of data, the Subspace Distribution Clustering Hidden Markov Model （SDCHMM）, derived from the Continuous Density Hidden Markov Model （CDHMM）, is introduced. With parameter tying, a new method to train SDCHMMs is described. Compared with the conventional training method, an SDCHMM recognizer trained by means of the new method achieves higher accuracy and speed. Experiment results show that the SDCHMM recognizer outperforms the CDHMM recognizer on speech recognition of Chinese digits.展开更多
文摘Clustering is one of the recently challenging tasks since there is an ever.growing amount of data in scientific research and commercial applications. High quality and fast document clustering algorithms are in great demand to deal with large volume of data. The computational requirements for bringing such growing amount data to a central site for clustering are complex. The proposed algorithm uses optimal centroids for K.Means clustering based on Particle Swarm Optimization(PSO).PSO is used to take advantage of its global search ability to provide optimal centroids which aids in generating more compact clusters with improved accuracy. This proposed methodology utilizes Hadoop and Map Reduce framework which provides distributed storage and analysis to support data intensive distributed applications. Experiments were performed on Reuter's and RCV1 document dataset which shows an improvement in accuracy with reduced execution time.
基金partially supported by the National Science Foundation of China (No.91738201,U21A20450 and 62171234)the Jiangsu Province Basic Research Project (No. BK20192002)the postgraduate research & practice innovation program of jiangsu province under Grant KYCX20_0708
文摘In this paper,we investigate the spectrum sensing performance of a distributed satellite clusters(DSC)under perturbation,aiming to enhance the sensing ability of weak signals in the coexistence of strong and weak signals.Specifically,we propose a cooperative beamforming(BF)algorithm though random antenna array theory to fit the location characteristic of DSC and derive the average far-field beam pattern under perturbation.Then,a constrained optimization problem with maximizing the signal to interference plus noise ratio(SINR)is modeled to obtain the BF weight vectors,and an approximate expression of SINR is presented in the presence of the mismatch of signal steering vector.Finally,we derive the closedform expression of the detection probability for the considered DSC over Shadowed-Rician fading channels.Simulation results are provided to validate our theoretical analysis and to characterize the impact of various parameters on the system performance.
基金the General Program of National Natural Science Foundation of China(62072049).
文摘The rapid growth of modern mobile devices leads to a large number of distributed data,which is extremely valuable for learning models.Unfortunately,model training by collecting all these original data to a centralized cloud server is not applicable due to data privacy and communication costs concerns,hindering artificial intelligence from empowering mobile devices.Moreover,these data are not identically and independently distributed(Non-IID)caused by their different context,which will deteriorate the performance of the model.To address these issues,we propose a novel Distributed Learning algorithm based on hierarchical clustering and Adaptive Dataset Condensation,named ADC-DL,which learns a shared model by collecting the synthetic samples generated on each device.To tackle the heterogeneity of data distribution,we propose an entropy topsis comprehensive tiering model for hierarchical clustering,which distinguishes clients in terms of their data characteristics.Subsequently,synthetic dummy samples are generated based on the hierarchical structure utilizing adaptive dataset condensation.The procedure of dataset condensation can be adjusted adaptively according to the tier of the client.Extensive experiments demonstrate that the performance of our ADC-DL is more outstanding in prediction accuracy and communication costs compared with existing algorithms.
基金supported by the Advanced Research Project of China Manned Space Program.
文摘The reconstruction of spacecraft cluster based on local information and distributed strategy is investigated.Each spacecraft is an intelligent individual that can detect information within a limited range and can determine its behavior based on surrounding information.The objective of the cluster is to achieve the formation reconstruction with minimum fuel consumption.Based on the principle of dual pulse rendezvous maneuver,three target selection strategies are designed for collision avoidance.Strategy-1 determines the target point’s attribution according to the target’s distance when the target point conflicts and uses a unit pulse to avoid a collision.Strategy-2 changes the collision avoidance behavior.When two spacecraft meet more than once,the strategy switches the target points of the two spacecraft.In Strategy-3,the spacecraft closer to the target has higher priority in target allocation.Strategy-3 also switches the target points when two spacecraft encounter more than once.The three strategies for a given position,different completion times,and random position are compared.Numerical simulations show that all three strategies can accomplish the spacecraft cluster's reconfiguration under the specified requirements.Strategy-3 is better than Strategy-1 in all simulation cases in the sense of less fuel consumption with different completion times and given location,and it is more effective than Strategy-2 in most of the completion time.With a random initial position and given time,Strategy-3 is better than Strategy-1 in about 70%of the cases and more stable.
文摘In many applications such as computational fluid dynamics and weather prediction, as well as image processing and state of Markov chain etc., the grade of matrix n is often very large, and any serial algorithm cannot solve the problems. A distributed cluster-based solution for very large linear equations is discussed, it includes the definitions of notations, partition of matrix, communication mechanism, and a master-slaver algorithm etc., the computing cost is O（n^3/N）, the memory cost is O（n^2/N）, the I/O cost is O（n^2/N）, and the com- munication cost is O（Nn ）, here, N is the number of computing nodes or processes. Some tests show that the solution could solve the double type of matrix under 10^6 × 10^6 effectively.
基金Supported by the Independent Scientific Research of Nanjing University of Science and Technology(2011YBXM110)~~
文摘The sequential dispersing results of aerial cluster bomb are discussed. The ballistic model and the mod- el for impact point distribution of bullets are established. The main factors influencing impact point distribution are analyzed by numerical simulation. And the feasibility of improving distribution effect through sequential dis- persing is validated. Sequential dispersion and optimized airdrop parameters can help to achieve better battle effec- tiveness.
基金This paper was supported by the National Science Foundation for Young Scholars of China under Crant No.61001091 .
文摘Wireless sensor network nodes （WSN nodes） have limited computing power, storage ca-pacity, conmmunication capabilities and energy and WSN nodes are easy to be paralyzed by Sybil at- tack. In order to prevent Sybil attacks, a new key distribution scheme for wireless sensor networks is presented. In this scheme, the key inforrmtion and node ID are associated, and then the attacker is dif-ficult to forge identity ID and the key inforrmtion corresponding to ID can not be forged. This scheme can use low-power to resist the Syhil attack and give full play to the resource advantages of the cluster head. The computing, storage and corrn^ni- cation is rminly undertaken by the cluster head o- verhead to achieve the lowest energy consumption and resist against nodes capture attack. Theoretical analysis and experimental results show that com- pared with the traditional scheme presented in Ref. [14], the capture rate of general nodes of cluster re-duces 40%, and the capture rate of cluster heads reduces 50%. So the scheme presented in this pa-per can improve resilience against nodes capture at- tack and reduce node power consumption.
基金supported by the National Natural Science Foundation of China(11671011).
文摘For a collective system,the connectedness of the adjacency matrix plays a key role in making the system achieve its emergent feature,such as flocking or multi-clustering.In this paper,we study a nonsymmetric multi-particle system with a constant and local cut-off weight.A distributed communication delay is also introduced into both the velocity adjoint term and the cut-off weight.As a new observation,we show that the desired multi-particle system undergoes both flocking and clustering behaviors when the eigenvalue 1 of the adjacency matrix is semi-simple.In this case,the adjacency matrix may lose the connectedness.In particular,the number of clusters is discussed by using subspace analysis.In terms of results,for both the non-critical and general neighbourhood situations,some criteria of flocking and clustering emergence with an exponential convergent rate are established by the standard matrix analysis for when the delay is free.As a distributed delay is involved,the corresponding criteria are also found,and these small time lags do not change the emergent properties qualitatively,but alter the final value in a nonlinear way.Consequently,some previous works[14]are extended.
基金Supported by the National Natural Science Foundation of China (No.60172048)
文摘As a kind of statistical method, the technique of Hidden Markov Model （HMM） is widely used for speech recognition. In order to train the HMM to be more effective with much less amount of data, the Subspace Distribution Clustering Hidden Markov Model （SDCHMM）, derived from the Continuous Density Hidden Markov Model （CDHMM）, is introduced. With parameter tying, a new method to train SDCHMMs is described. Compared with the conventional training method, an SDCHMM recognizer trained by means of the new method achieves higher accuracy and speed. Experiment results show that the SDCHMM recognizer outperforms the CDHMM recognizer on speech recognition of Chinese digits.
