Entity

Time filter

Source Type


Shen Q.,Jinling Institute of Technology | Shen Q.,Jiangsu Information Analysis Engineering Laboratory
Applied Mechanics and Materials | Year: 2014

In the field of artificial intelligence, Case Based Reasoning (CBR) is an arising reasoning technique, in which choosing reasoning index has been the hot topic and difficulty. In order to get optimal feature subset in the process of index selection, this paper combined gray correlation analysis with genetic algorithm (GA) to optimize the feature selection process, taking the gray correlation analysis result as the initial population for GA heuristic search, which can on one hand get better feature combination, on the other hand, effectively reduce the evolution of GA, finally improving GA's execution efficiency. Based on this, an optimized GA-CBR case reasoning model is put forward, which has been proved by empirical results to have an improved CBR forecasting accuracy. © (2014) Trans Tech Publications, Switzerland. Source


Shao F.,Jiangsu Information Analysis Engineering Laboratory | Shao F.,Nanjing University of Posts and Telecommunications | Shao F.,Jinling Institute of Technology | Jiang G.,Nanjing University of Posts and Telecommunications
Advances in Information Sciences and Service Sciences | Year: 2012

To investigate the effect of the community structure on the traffic-driven epidemic spreading in homogeneous networks, a model of pseudo-random network is presented with adjustable community structure and constant average degree. A novel and effective control strategy for controlling epidemic spreading is proposed based on the model. By adding edges in proper order according to their weights, which are defined in different ways, the propagation velocity and the epidemic threshold are changed. The control strategy by adding edges between nodes whose product of node degrees is highest is proven to be the most efficient. It can reduce the propagation velocity in SI model and improve the epidemic threshold in SIS model, especially in networks with pronounced community structure. Simulation results have confirmed the theoretical predictions. Source


Shao F.,Jiangsu Information Analysis Engineering Laboratory | Shao F.,Nanjing University of Posts and Telecommunications | Shao F.,Jinling Institute of Technology | Jiang G.P.,Nanjing University of Posts and Telecommunications
Journal of Networks | Year: 2012

To understand the effect of the community structure on the epidemic propagation in homogeneous networks, a model of pseudo-random network is presented with adjustable community structure and constant average degree. In the scenario that the propagation is driven by reaction processes from nodes to all neighbors, pronounced community structure can reduce the epidemic propagation velocity. While in the situation that epidemic pathway is defined by traffic flows, the epidemic spreading in networks with pronounced community structure is obviously accelerated instead. When it is extended to the SIS model with traffic flows, the epidemic threshold is found to be proportional to the inverse of the average betweenness. Simulation results have confirmed the theoretical predictions. © 2012 ACADEMY PUBLISHER. Source


Chang Z.-N.,Jinling Institute of Technology | Chang Z.-N.,Jiangsu Information Analysis Engineering Laboratory
2011 2nd International Conference on Artificial Intelligence, Management Science and Electronic Commerce, AIMSEC 2011 - Proceedings | Year: 2011

Datasets used in financial distress forecast are unbalanced. The traditional method gets lower predict accuracy especially in small samples of unbalanced datasets. The datasets are balanced with SMOTE method and then classified with the classical decision tree algorithm C4.5. The results show that the prediction model based on C4.5 algorithm gets the better performance. © 2011 IEEE. Source


Shen Q.,Jiangsu Information Analysis Engineering Laboratory | Shen Q.,Jinling Institute of Technology | Chen A.,Jinling Institute of Technology
Proceedings - 2012 IEEE Symposium on Robotics and Applications, ISRA 2012 | Year: 2012

Similar case retrieval ability is a key technology in CBR(Case Based Reasoning) system. In order to improve the case retrieval efficiency in business financial distress warning(FDW) system, a CBR case retrieval model based on gray relation was proposed, applying the gray relational analysis in case based reasoning for business FDW which improved the deficiency of distance measurement. Moreover, taking into account the different importance of case features in predicting financial distress, a weight vector was defined to solve the influence coming from nor-crucial features. Empirical results proves that this method can effectively improve the case retrieval efficiency of the target enterprise in business FDW system. © 2012 IEEE. Source

Discover hidden collaborations