Detecting DDoS attacks using a cascade of machine learning classifiers based on Random Forest and MLP-ANN

Trevor Pinto, Yakub Sebastian

Research output: Chapter in Book/Report/Conference proceedingConference Paper published in Proceedingspeer-review


Distributed Denial of Service (DDoS) attacks are one of the most versatile and powerful cyber-attack that has remarkably grown in sophistication, throughput as well as bandwidth. In addition to multiple efficient, effective, and accurate ML approaches been proposed over the years, DDoS detection remains a challenge due to ever-growing advancements in network technology. In this paper, a hybrid approach incorporating machine learning algorithms towards the detection of DDoS attacks is proposed. This approach includes a classifier that is formed by cascading two machine learning algorithms, Random Forest (RF) with a Multi-layer Perceptron (MLP) Neural Network. These algorithms were chosen due to their high real-time accuracies, efficiency, adaptability, and flexibility in changing the parameters which is vital due to the complex and changeable nature of network technology, that may cause inaccurate results if a set of features turn obsolete over time. Feature selection is done carefully to obtain the highest possible accuracy and efficiency through Information Gain (IG) algorithm which would aid towards detection of modern cyber-attacks including zero-day attacks.
Original languageEnglish
Title of host publication2021 IEEE Madras Section Conference (MASCON)
Place of PublicationPiscataway, NJ
PublisherIEEE, Institute of Electrical and Electronics Engineers
Number of pages7
ISBN (Electronic)978-1-6654-0405-1
ISBN (Print)978-1-6654-4736-2
Publication statusPublished - 27 Aug 2021
Event2021 IEEE Madras Section Conference - Chennai, India
Duration: 27 Aug 202128 Aug 2021


Conference2021 IEEE Madras Section Conference
Abbreviated titleMASCON


Dive into the research topics of 'Detecting DDoS attacks using a cascade of machine learning classifiers based on Random Forest and MLP-ANN'. Together they form a unique fingerprint.

Cite this