Hide menu

Examensarbeten och uppsatser / Final Theses

Framläggningar på IDA / Presentations at IDA


Se även framläggningar annonserade hos ISY och ITN i Norrköping / See also presentations announced at ISY and ITN in Norrköping (in Swedish)

If nothing is stated about the presentation language then the presentation is in Swedish.

WExUpp - kommande framläggningar
2019-08-19 - SaS
Implementations of Principal Component Analysis in CUDA for Real-time Anomaly Detection
Joakim Bertils
Avancerad (30hp)
kl 08:15, Alan Turing (In English)
2019-08-20 - SaS
Defining new machine learning efficiency metrics
Anton Dalgren, Ylva Lundegård
Avancerad (30hp)
kl 13:15, Alan Turing (In English)
[Abstract]
Impressive results can be achieved when stacking deep neural networks hierarchies together. Several machine learning papers claim state-of-the-art results when evaluating their models with different accuracy metrics. However, these models come at a cost, which is rarely taken into consideration. This thesis aims to shed light on the resource consumption of machine learning, and therefore, five efficiency metrics is proposed. These should be used for evaluating machine learning models, taking accuracy, model size, and time and energy for both training and inference into account. These metrics are intended to make a fairer evaluation of machine learning models, not only looking at accuracy. This thesis presents an example of how these metrics can be used by applying them to both text and image classification tasks using the algorithms SVM, MLP, and CNN.
2019-08-22 - SaS
Evaluation of two vulnerability scanners accuracy and consistency in a cyber range
Erik Hyllienmark
Avancerad (30hp)
kl 10:00, Alan Turing (In English)
[Abstract]
One challenge when conducting exercises in a cyber range is to know what applications and vulnerabilities are present on deployed computers. In this paper, the reliability of application- and vulnerability reporting by two vulnerability scanners, OpenVas and Nexpose, have been evaluated based on their accuracy and consistency. Followed by an experiment, the configurations on two virtual computers were varied in order to identify where each scanner gathers information. Accuracy was evaluated with the f1-score, which combines the precision and recall metric into a single number. Precision and recall were calculated by comparing installed applications and vulnerabilities on virtual computers with the scanning reports. Consistency was evaluated by quantifying how similar the reporting of applications and vulnerabilities between multiple vulnerability scans were into a number between 0 and 1. The vulnerabilities reported by both scanners were also combined with their union and intersection to increase the accuracy. The evaluation reveal that neither Nexpose or OpenVas accurately and consistently report installed applications and vulnerabilities. Nexpose reported vulnerabilities better than OpenVas with an accuracy of 0.78. Nexpose also reported applications more accurately with an accuracy of 0.96. None of the scanners reported both applications and vulnerabilities consistently over three vulnerability scans. By taking the union of the reported vulnerabilities by both scanners, the accuracy increased by 8 percent compared with the accuracy of Nexpose alone. However, our conclusion is that the scanners' reporting does not perform well enough to be used for a reliable inventory of applications and vulnerabilities in a cyber range

Page responsible: Ola Leifler
Last updated: 2017-04-27