| Grant number: | 11/14058-5 |
| Support type: | Scholarships in Brazil - Master |
| Effective date (Start): | March 01, 2012 |
| Effective date (End): | August 31, 2013 |
| Field of knowledge: | Physical Sciences and Mathematics - Computer Science |
| Principal Investigator: | João Paulo Papa |
| Grantee: | Rodrigo Yuji Mizobe Nakamura |
| Home Institution: | Faculdade de Ciências (FC). Universidade Estadual Paulista (UNESP). Campus de Bauru. Bauru , SP, Brazil |
| Associated research grant: | 09/16206-1 - New trends on optimum-path forest-based pattern recognition, AP.JP |
Abstract Traditional pattern recognition techniques assume that data is independent and identically distributed. Although in many real cases this is not true, in the context of intrusion detection, for instance, each sample of the dataset represents an access to a particular computer. The feature vector is created using TCP/IP packet information, such as source and destination address, packet size and permission to perform some process. Henceforth, it is necessary multiple accesses to the same machine to be classified as a particular attack. Thus, analyzing every individual sample may not be a very interesting alternative. In order to tackle this problem, sequential supervised learning techniques use spatial and/or temporal dependence to increase the effectiveness of traditional approaches. Among the many conventional ones, Optimum-Path Forest (OPF) was recently proposed in order to ally efficiency in the training phase and effectiveness in classification. Therefore, the present research work aims to study some sequential learning approaches and then to implement them with OPF. The main contribution of this work is to implement OPF classifier in the sequential supervised learning environment, since this task has not been done so far. | |