The Machine Learning theme consolidates a broad range of research activities in Informatics that are related to machine learning. The activities are related to methodology (models and algorithms), systems, as well as applications, and they span several areas including artificial intelligence, biomedical/health informatics, communications, computer vision, cybersecurity, data analysis, and robotics. Broad questions that have or are being addressed in the context of machine learning include:  how to classify or cluster objects into meaningful categories, how to detect actionable patterns or extract relationships from data, how to explain the output of machine learning to users, how to predict values or trends from data, and how to recommend actions to users.

 

Key research activities in the area of Machine Learning:

• Accountability and explainability of machine learning models
• Adversarial machine learning
• Automated reasoning processes through machine learning/data mining models
• Automatic relationship extraction from unstructured data and for document classification
• Clustering algorithms for data privacy protection
• Computer vision through neural networks, deep learning, sparse dictionary learning
• Communication of Narrowband Internet-of-Things devices and UAVs and resource allocation problems in full-duplex UAV-enabled communication networks
• Control engineering applications for robotics
• Detection of buried landmines in the ground through classification and prediction of landmine risk
• Development of  kernel methods, active search, statistical learning theory, large-scale learning methods, Bayesian optimization methods, genetic algorithms
• Discrimination in machine learning
• Disease classification and patient stratification
• EMG/ECG/Bio-signal classification
• Extraction of relationships from unstructured data
• Hand motions and muscle activity classification
• Human movement understanding in the context of robotics
• Machine learning in teaching
• Pattern mining algorithms for sequential data and data streams
• Prediction of content that will be watched in BBC player, bias in social media, user engagement
• Recommendation through randomized graph mining processes
• Recommendation of social media content
• Security and privacy of machine learning in the context of social media
• Security through classification of malicious behaviours, concept drift detection in security tasks, frameworks based on supervised, online, and active learning, and evaluation of machine learning tasks in security contexts.
• Text document classification