Litcius/Paper detail

XSS Filter detection using Trust Region Policy Optimization

Biswajit Mondal, Abhijit Banerjee, Subir Gupta

202310 citationsDOI

Abstract

Cross-site scripting (XSS)has gotten little attention regarding detecting and keeping it secure, leaving artificial intelligence systems susceptible to assault. It is crucial to determine ways to make the detecting system more attack-resistant. This study aims to employ Trust Region Policy optimization (TRPO) reinforcement learning techniques to enhance XSS detection and prevent adversarial attacks. Before mining the model’s hostile inputs, the model’s information is obtained using a reinforcement learning framework. Second, a detection method and an adversarial method are simultaneously trained. New damaging data is introduced to the detection model every cycle to retrain it. The proposed XSS model mines risky inputs that black-box or white-box detection approaches miss during testing. It has been proved that the escape rate can be decreased by simultaneously training the detection technique and the attack model. It increases the models’ capacity for self-defense.

Topics & Concepts

Cross-site scriptingComputer scienceReinforcement learningScripting languageAdversarial systemFilter (signal processing)Computer securityArtificial intelligenceMachine learningComputer visionWorld Wide WebWeb application securityThe InternetWeb developmentOperating systemAdvanced Malware Detection TechniquesDigital and Cyber ForensicsWeb Application Security Vulnerabilities