Optimal state and branch sequence based parameter estimation of continuous hidden Markov model

A parameter estimation algorithm of the continuous hidden Markov model isintroduced and the rigorous proof of its convergence is also included. The algorithm uses theViterbi algorithm instead of K-means clustering used in the segmental K-means algorithm to determineoptimal state and branch sequences. Based on the optimal sequence, parameters are estimated withmaximum-likelihood as objective functions. Comparisons with the traditional Baum-Welch and segmentalK-means algorithms on various aspects, such as optimal objectives and fundamentals, are made. Allthree algorithms are applied to face recognition. Results indicate that the proposed algorithm canreduce training time with comparable recognition rate and it is least sensitive to the training set.So its average performance exceeds the other two.

Subspace Distribution Clustering HMM for Chinese Digit Speech Recognition

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.

Research on anti-attack performance of a private cloud safety computer based on the Markov-Percopy dynamic heterogeneous redundancy structure

With the increasing computing demand of train operation control systems,the application of cloud computing technology on safety computer platforms of train control system has become a research hotspot in recent years.How to improve the safety and availability of private cloud safety computers is the key problem when applying cloud computing to train operation control systems.Because the cloud computing platform is in an open network environment,it can face many security loopholes and malicious network at-tacks.Therefore,it is necessary to change the existing safety computer platform structure to improve the attack resistance of the private cloud safety computer platform,thereby enhancing its safety and reliability.Firstly,a private cloud safety computer platform architecture based on dynamic heterogeneous redundant(DHR)structure is proposed,and a dynamic migration mechanism for heterogeneous executives is designed.Then,a generalized stochastic Petri net(GSPN)model of a private cloud safety computer platform based on DHR is established,and its steady-state probability is solved by using its isomorphism with the continuous-time Markov model(CTMC)to analyse the impact of different system structures and executive migration mechanisms on the system's anti-attack performance.Finally,through experimental verifcation,the system structure proposed in this paper can improve the anti-attack capability of the private cloud safety computer platform,thereby improving its safety and reliability.

Research on the logic monitoring method for the cloud computing based safety computer

With the development of railway construction in China,the computing demand of the train control system is increasing day by day.The application of cloud computing technology on the rail transit signal system has become a research hotspot in recent years.How to improve the safety and availability of the safety computer platform in the cloud computing environment is the key problem when applying cloud computing to the train operation control system.Since the cloud platform is in an open network environment,fac-ing many security vulnerabilities and malicious network attacks,it is necessary to monitor the operation of computer programmes through edge safety nodes.Firstly,this paper encrypts the logical monitoring method,and then proposes a secure computer de fence model based on the dynamic heterogeneous redundancy structure.Then the continuous time Markov chain(CTMC)is used to quantitatively solve the stable probability of the system,and the influence of different logical monitoring methods on the anti-attack performance of the system is analysed.Finally,the experiment proves that the dynamic heterogeneous redundancy structure composed of encryption logic monitoring can guarantee the safe and stable operation of the safety computer more effectively.