Machine Learning Group

Machine learning (ML) enables computers to learn from data and use what they learn to make predictions, support decisions, and perform complex tasks. These techniques are central to artificial intelligence (AI) and data science.

Machine learning is useful when the patterns in data are too complex, numerous, or changeable to be captured by handwritten rules. Instead of specifying exactly how every case should be handled, we can provide examples and allow an algorithm to learn a model from them.

Consider an email filter. A fixed collection of rules can quickly become outdated as spammers change their tactics. A machine learning system can instead learn from examples of spam and legitimate email, identifying combinations of characteristics that distinguish them. Similar methods can uncover associations in shopping data, recognise objects in images, analyse language, detect unusual events, and forecast future outcomes.

Machine learning algorithms must be accurate, efficient, and robust, and their results must be evaluated carefully. The algorithms and their outputs should also be accessible to the people who understand the problems and the contexts in which the technology will be used. The Waikato Machine Learning Group develops improved algorithms for machine learning and applies them to practical problems in Aotearoa and beyond.

Machine learning research at Waikato has a long history, rooted in practical applications. Waikato’s open-source machine learning software, including the Waikato Environment for Knowledge Analysis (WEKA workbench), first publicly released in 1996, was originally developed with particular attention to problems in agriculture and horticulture. Waikato researchers have applied machine learning in areas including dairy herd management, bioinformatics, retail data, soil analysis, and natural language processing.

This combination of fundamental research, accessible software, and practical impact continues to shape the group’s work.

Further details on Machine Learning Group's website.