A brief overview of machine learning: What? How? When? … and should I care anyway?
Bogdan J. Matuszewski (University of Central Lancashire, UK)
With continued advances in mathematical modelling, ever increasing computational power and the recent unprecedented explosion of shared information (e.g. with reported hundreds of hours of video uploaded to the YouTube servers every minute) or creation of large databases most, notably in biomedicine or astronomy (e.g. consider Pan-STARRS 1.6 petabytes data) the growth in automated data analysis is unavoidable. Without the deployment of machine learning techniques, handling rapidly growing data volumes may not be possible.
The talk will focus on the recent advances and fundamentals the of machine learning with key ideas and terminology explained. It will include machine learning taxonomy and explain briefly relevant concepts of artificial intelligence, data mining, pattern recognition and computer vision. The focus will be on a broad overview of key methodologies, including: supervised, semi-supervised, unsupervised, active and reinforced-learning and their use in classification and regression problems. Essential concepts of: training, validation and testing; evaluation methodologies and metrics; as well as fundamental limitations and frequent misconceptions will be mentioned. The talk will include some historical background as well a discussion of the recent state-of-the-art. A representative sample of machine learning techniques will be briefly explained, including deep learning.
A small number of practical examples, including biomedical and Industry 4.0 applications, will be provided to succinctly illustrate the key machine learning concepts.