Applying the rainbow architecture to intrusion detection systems

There is a lot of expectation on how Artificial Intelligence (AI) is going to have an impact on Cybersecurity. From new sophisticated attacks to new ways of defending a system from cybercriminals. A lot of techniques are being studied by universities and security organizations in order to implement...

Descripción completa

Detalles Bibliográficos
Autor: Izquierdo García-Faria, Tomás
Tipo de recurso: tesis de maestría
Fecha de publicación:2021
País:España
Institución:Universitat Politècnica de Catalunya (UPC)
Repositorio:UPCommons. Portal del coneixement obert de la UPC
Idioma:inglés
OAI Identifier:oai:upcommons.upc.edu:2117/348194
Acceso en línea:https://hdl.handle.net/2117/348194
Access Level:acceso abierto
Palabra clave:Computer security
Deep learning
Reinforcement learning
Ciberseguretat
Aprenentatge profund de reforç
Sistemes de detecció d'intrusos
IDS
Aprenentatge de reforç
Aprenentatge Q
Rainbow Aprenentatge automàtic
Cybersecurity
Deep Reinforcement Learning
Intrusion Detection Systems
Reinforcement Learning
Q Learning
Rainbow
Machine Learning
Seguretat informàtica
Aprenentatge profund
Aprenentatge per reforç
Àrees temàtiques de la UPC::Informàtica::Intel·ligència artificial
Descripción
Sumario:There is a lot of expectation on how Artificial Intelligence (AI) is going to have an impact on Cybersecurity. From new sophisticated attacks to new ways of defending a system from cybercriminals. A lot of techniques are being studied by universities and security organizations in order to implement intelligent systems to defend systems, as well as researching how can AI give solution to many problems not yet solved. However this sector has some limitations for AI to have a real impact. AI is a discipline which, in its machine learning part, needs data in order to give good results and big companies are not willing to admit there attacks or do not even know they are under attack. In particular network intrusion is one of the most common ways attackers enter a system and exploit its threats. With the adoption of IoT systems in our society more and more, companies, cities, and citizens, are going to fill their surroundings with devices that communicate to the exterior. Under these emerging threats every system and network they are connected to have to be secure and should be able to detect intrusion. This thesis will try to present a conjunction of different techniques, presented by their authors as the Rainbow architecture, used in reinforcement learning in order to tackle this problem. It will give solution to systems such as intrusion detection systems (IDS) and can be extended to intrusion prevention systems (IPS) or web application firewalls (WAF).