Machine Learning

The aim of the group is to investigate automated learning in intelligent systems by developing computational models and algorithms. This group is led by Professor Danushka Bollegala.

The Machine Learning (ML) group in the Department of Computer Science includes academics, research staff and PhD students specialising in all aspects of machine learning and data analysis techniques.

Specifically, the group consists of experts that focus on these core areas.

Natural language processing

Aims to provide principled solutions and is applied across a wide range of areas including automated sentiment classification and social media analysis.

Multi-agent Reinforcement Learning

Focuses on problems of coordination and cooperation in multi-agent systems. Aims to develop new multi-agent reinforcement learning algorithms that are more sample-efficient, robust, and scalable. 

Data mining and analysis

Includes time series analysis authentication and mechanisms for achieving secure data mining over encrypted data. The group members have worked on various types of data sources including but not limited to health, financial, legal and e-commerce.

Human-AI interaction

Aims to develop new interactive learning systems that allow non-expert humans to teach agents new complex tasks easily and effectively, and to enable agents to learn more efficiently from human teachers. 

Combinatorial Data Analytics

We investigate various combinatorial models for object sequencing and ordering as well as measures for evaluating such sequences, and also relaxation based optimisation techniques for large scale optimisation of various combinatorial models.

Representation learning

The conducted research designs effective ways to automatically discover from raw data suitably informative representations for machine learning tasks and also for data visualisation. Models are built on stochastic, spectral embedding and projection techniques, neural networks, information theory, matrix factorisation and constrained optimisation.Performance-sensitive Machine Learning

Performance-sensitive Machine Learning

Performance-sensitive Machine Learning covers the scenarios where a machine learning model is trained or deployed in a hardware environment where the hardware limitations are the primary concerns when designing a solution. This covers training or inference of Deep networks on mobile devices, as well as programs combining Deep Learning with traditional numerical solvers involving Partial-Differential Equations. 

 

For the schedule of ML group meetings and talks please see the research page.

 

People

Academic staff members of this group are:

 

Back to: Department of Computer Science