Banca de DEFESA: Rafael Bruno Peccatiello

Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.
STUDENT : Rafael Bruno Peccatiello
DATE: 01/08/2023
TIME: 08:00
LOCAL: https://teams.microsoft.com/l/meetup-join/19%3ameeting_NWE3YzY3OTAtNWRjZS00ODE3LWEzOTAtODM1YjFlMmNjZ
TITLE:

Detection of malicious behavior of internal users on networks using data flow analysis

KEY WORDS:

Insider Threat, Data Stream, Machine Learning, Unbalance

PAGES: 51
BIG AREA: Ciências Exatas e da Terra
AREA: Ciência da Computação
SUMMARY:

The present study addresses the problem of detecting insider threats. An insider threat is anyone who has legitimate access to a particular organization's network and uses that access to harm that organization. The study presents a way of solving the problem through the proposition of a model that brings together different Data Science techniques. Among them are the use of supervised Machine Learning and the use of the data flow analysis approach. The latter was developed with the application of a way for the treatment of unbalanced data flows, since the unbalance is a variable present in the domain of the problem. The unbalance of the flow will be treated through propagation of minority samples for balanced composition of a sliding training window. The attributes and the way in which they were extracted come from sources present in the literature. The algorithms used were Random Forest (RF), Light Gradient Boosting Machine (LGBM) and K-Nearest Neighborhoods (KNN). The performance of the model was evaluated according to the values obtained by the metrics precision, recall, F1-Score and kappa. The results obtained show that the treatment of unbalance allows achieving superior performance to those obtained in the literature, for insider threat detection.

COMMITTEE MEMBERS:
Presidente - 3128249 - LUIS PAULO FAINA GARCIA
Interno - 2518570 - MARCOS FAGUNDES CAETANO
Externo à Instituição - SYLVIO BARBON JÚNIOR