No description available.

No description available.

No description available.

Standard Markov chain Monte Carlo (MCMC) sampling methods can experience problem sampling from multi-modal distributions. A variety of sampling methods have been introduced to overcome this problem. The mode jumping method of Tjelmeland & Hegstad (2001) tries to find a mode and propose a value from that mode in each mode jumping attempt. This approach is inefficient in that the work needed to find each mode and model the distribution in a neighbourhood of the mode is carried out repeatedly during the sampling process. We shall propose a new mode jumping approach which retains features of the Tjelmeland & Hegstad (2001) method but differs in that it finds the modes in an initial search, then uses this information to jump between modes effectively in the sampling run. Although this approach does not allow a second chance to find modes in the sampling run, we can show that the overall probability of missing a mode in our approach is still low. We apply our methods to sample from distributions which have continuous variables, discrete variables, a mixture of discrete and continuous variables and variable dimension. We show that our methods work well in each case and in general, are better than the MCMC sampling methods commonly used in these cases and also, are better than the Tjelmeland & Hegstad (2001) method in particular.

This thesis focuses on the search for neutrinos from the solar hep reaction using the combined three phases of the Sudbury Neutrino Observatory (SNO) data. The data were taken over the years 19992006, totalling 1,083 days of live neutrino time. The previous published SNO hep neutrino search was completed in 2001 and only included the first phase of data taking. That hep search used an event counting approach in one energy bin with no energy spectral information included. This thesis will use a spectral analysis approach. The hep neutrino search will be a Bayesian analysis using Markov Chain Monte Carlo (MCMC), and a Metropolis-Hastings algorithm to sample the likelihood space. The method allows us to determine the best fit values for the parameters. This signal extraction will measure the 8B flux, the atmospheric neutrino background rate in the SNO detector, and the hep flux. This thesis describes the tests used to verify the MCMC algorithm and signal extraction. It defines the systematic uncertainties and how they were accounted for in the fit. It also shows the correlations between all of the parameters and the effect of each systematic uncertainty on the result. The three phase hep signal extraction was completed using only 1/3 of the full data set. With these lowered statistics, this analysis was able to place an upper limit on the hep flux of 4.2 10^4 cm2 s1 with a 90% confidence limit. It was able to measure a hep flux of (2.40(+1.19)(-1.60))10^4 cm2 s1. These numbers can be compared with the previous SNO upper limit of 2.310^4 cm2 s1 with a 90% confidence limit, and the standard solar model prediction of (7.970 1.236) 10^3 cm2 s1.

none

We present a new detection model include monitoring network perimeter and hosts logs to counter the new method of attacking involve different hosts source during an attacking sequence. The new attacking sequence we called ¡§Scout and Intruder¡¨ involve two separate hosts. The scout will scan and evaluate the target area to find the possible victims and their vulnerability, and the intruder launch the precision strike with login activities looked as same as authorized users. By launching the scout and assassin attack, the attacker could access the system without being detected by the network and system intrusion detection system. In order to detect the Scout and intruder attack, we correlate the netflow connection records, the system logs and network data dump, by finding the states of the attack and the corresponding features we create the detection model using the Hidden Markov Chain. With the model we created, we could find the potential Scout and the Intruder attack in the initial state, which gives the network/system administrator more response time to stop the attack from the attackers.

A wireless sensor network can be used to track an object. Every sensor has limited energy and detecting range. In order to conserve energy, sensors may be put into sleeping mode. A sensor in the sleeping mode can not communicate with other sensors or detect objects. When the object moves to the sensing range of a sleeping sensor, a tracking error occurs. To minimize the tracking error subject to an constraint on energy consumption, we should determine the sleeping schedules of sensors based on the mobility pattern of the object. We propose determining the sleeping schedules based on the observation history of the moving object. We use computer simulation to justify the usage of the proposed approach.

No description available.

No description available.