Cui Y.-Y.,Southwest Jiaotong University |
Wang J.-Q.,Southwest Jiaotong University |
Zhuang H.-G.,Shanghai Stock Exchange
Xitong Gongcheng Lilun yu Shijian/System Engineering Theory and Practice | Year: 2011
Given the limitation of traditional portfolio theory, portfolio model using a Bayesian decision theory and skew-normal distributions was proposed to solve the problem of parameter uncertainty and improve utility of portfolio. Then, based on Taylor series expansion of expected utility, we analysed the relationship between the higher moment risk and utility, and employed the MCMC and numerical optimization algorithm which estimated the parameters of skew-normal distributions and the weights of portfolio. Our results suggest that expected utility can be improved using Bayesian Theory which solves the problem of parameter uncertainty. Further, it is important to incorporate mean, variance and skewness in portfolio selection strategy.
Zhang D.,Beijing Jiaotong University |
Jin D.,Beijing University of Posts and Telecommunications |
Gong Y.,Beijing University of Posts and Telecommunications |
Chen S.,Inner Mongolia University |
Wang C.,Shanghai Stock Exchange
Tehnicki Vjesnik | Year: 2015
Traditional static defect detection tools can detect software defects and report alarms, but the correlations among alarms are not identified and massive independent alarms are against the understanding. Helping users in the alarm verification task is a major challenge for current static defect detection tools. In this paper, we formally introduce alarm correlations. If the occurrence of one alarm causes another alarm, we say that they are correlated. If one dominant alarm is uniquely correlated with another, we know verifying the first will also verify the others. Guided by the correlation, we can reduce the number of alarms required for verification. Our algorithms are inter-procedural, path-sensitive, and scalable. We present a correlation procedure summary model for inter-procedural alarm correlation calculation. The underlying algorithms are implemented inside our defect detection tools. We chose one common semantic fault as a case study and proved that our method has the effect of reducing 34,23% of workload. Using correlation information, we are able to automate the alarm verification that previously had to be done manually. © 2015, Strojarski Facultet. All rights reserved.
Cui B.,Shandong University of Science and Technology |
Wang H.,South University of Science and Technology of China |
Ye K.,Shanghai Stock Exchange |
Yan J.,City University of Hong Kong
Expert Systems with Applications | Year: 2012
Agent-based computational economics (ACE) has received increased attention and importance over recent years. Some researchers have attempted to develop an agent-based model of the stock market to investigate the behavior of investors and provide decision support for innovation of trading mechanisms. However, challenges remain regarding the design and implementation of such a model, due to the complexity of investors, financial information, policies, and so on. This paper will describe a novel architecture to model the stock market by utilizing stock agent, finance agent and investor agent. Each type of investor agent has a different investment strategy and learning method. A prototype system for supporting stock market simulation and evolution is also presented to demonstrate the practicality and feasibility of the proposed intelligent agent-based artificial stock market system architecture. © 2012 Elsevier Ltd. All rights reserved.
Lei X.,Tongji University |
Yang X.,Tongji University |
Qi Z.,Shenhua Hollysys Information Technology Co. |
Lili L.,Shanghai Stock Exchange
Proceedings - 2012 3rd Global Congress on Intelligent Systems, GCIS 2012 | Year: 2012
Traditional k-nearest neighborhood (KNN) model is being widely used in the recommender systems. However, it behaves badly without enough history records for new users, called the cold starting problem. Both time and space complexity are huge for computing all pairwise similarities among items or users. A mixed neighborhood algorithm is proposed for treating new users and old users separately. For new users, this paper takes into account users' characteristics. For old users, combined with Singular Value Decomposition (SVD), we reduce the time and space complexity efficiently. Experiment on MovieLens dataset shows that the proposed model can solve the cold starting problem in effect and remarkably improve the accuracy of traditional model and lower time consuming level. © 2012 IEEE.
Chen Q.,Tongji University |
Xiang Y.,Tongji University |
Zhang X.,Shanghai Stock Exchange |
Guo X.,Tongji University |
Wang P.,Tongji University
International Journal of Advancements in Computing Technology | Year: 2012
Text Stream Mining (TSM) is concerned with mining and discovering temporal dynamic latent patterns in a stream of text such as research literature and newswire which application system collected over time, which is very useful in navigating and organizing corpus, as well as tracking the latent topic trends. The author focused mainly on the subtask of TSM called Evolutionary Pattern Mining (EPM) in which topic life cycle graph are constructed to help us gain intuitive sense on various topic properties in text stream. In this paper, topic related definitions in text stream were made formally, in which topics owned many properties e.g. popularity, attention impact, etc. Then topic ontology trees were extracted and constructed using nonparametric probabilistic Bayesian model. Finally topic life cycle curve was incrementally constructed using infinite Hidden Markov model (iHMM). Compared with LDA topic models, it doesn't need to specify the total number of topics in advance to achieve the purpose of text dimension reduction. Experiments on research literature dataset showed that the proposed method in this paper can efficiently discover and intuitively demonstrate interesting life cycle of certain property of a topic.