Anomaly detection is a concept widely applied to numerous domains, such as cyber intrusion, financial fraud, and many others. Within this topic, the student will focus on development of algorithms of anomaly detection using machine learning. The proposed principles will be applied mainly on text data (normal text, network syslogs, etc.) to automatically identify anomalies in large datasets. Some of the fields, where the developed anomaly detection algorithms will be applied on, include e-mails, network performance syslogs, network security syslogs and patient symptoms data. Student will primarily consider supervised and unsupervised machine learning concepts, as well as deep learning techniques. The developed algorithms will be verified through the numerical simulations as well as implementation in experimental environment. Preferred programming language to work with is PYTHON.

More specifically, the master thesis will deal with applying numerous machine learning models on many types of datasets. The objective is to find anomalies in data based on past dataset. The student is expected to analyze and compare the results and find the optimum models.

- Applying multiple machine learning models on various time series dataset to predict the future.
- Analyze the results and find the optimum models.
- Evaluate the proposed model and compare with other in the market.