Oppgaven er ikke lenger tilgjengelig

Identification and Classification of Threats using Neural Networks

The idea is based on a system of separately trained neural networks that perform different tasks to find, identify and classify threats through any sort of initial input.

The system is designed with cascade model in mind. The "first responder" neural network will be trained to find a point of an anomaly in supplied data and transfer the necessary information to the deeper layers where more detailed analysis will be performed. When the system is done with classifying, a report is generated by what kind of tasks system is tuned to.

Emneord: Neural Networks, Machine Learning, Deep Learning, Insider Threats, External Threats
Publisert 23. nov. 2017 13:25 - Sist endret 23. nov. 2017 13:36


Omfang (studiepoeng)