Presentation is loading. Please wait.

Presentation is loading. Please wait.

Intrusion Detection with Neural Networks my awesome graphic ↑

Similar presentations


Presentation on theme: "Intrusion Detection with Neural Networks my awesome graphic ↑"— Presentation transcript:

1 Intrusion Detection with Neural Networks my awesome graphic ↑
Mary Krajnak

2 Problem Definition Computers an essential part of our lives
Security has become an essential issue Methods of hand-coded rules for detecting misuse are not reliable and are labor intensive Need an intrusion detection system which works by modeling “normal behavior” to detect anomalies.

3 Data & ANN Structure DARPA Intrusion Detection Evaluation Session:
Collection of controlled TCP/IP connections which model normal network behavior Session: <Session ID, Start date, Start time, Duration, Service, Src port, Dest port, Src IP, Dest IP, Attack score, Attack name> Backpropagation Neural Network Data most likely not linearly seperable

4 Initial Work and Results
Extracted features from TCP/IP dump Created program to extract & convert Created a Backpropagation neural network in F90 Speed important in network security Initial results

5 Work To Be Done Pre-process data to obtain better results
Currently using four features <duration, service, source port, destination port> Feature reduction/addition may improve Implement C++ Backpropagation algorithm Speed comparable, may make up for authors penchant for mistakes in Fortran 90


Download ppt "Intrusion Detection with Neural Networks my awesome graphic ↑"

Similar presentations


Ads by Google