Semi-supervised learning paper selected as 2020 NatureSpringer Research Highlight
An article that Holger Hoos, Professor of Machine Learning at the Leiden Institute of Advanced Computer Science (LIACS), wrote with his former Master student Jasper van Engelen has been selected by SpringerNature as one of six Research Highlights in Computer Science of 2020. After becoming an ACM Fellow in January, it is another achievement Hoos can now add to his ongoing list of accomplishments.
Out of the 65,000 articles SpringerNature published in the field of Computer Science in 2020, they selected the survey on semi supervised learning, written by Van Engelen and Hoos, along with 5 other papers. ‘Semi-supervised learning is a fascinating but rather underappreciated area of machine learning with vast application potential,’ Hoos explains. ‘In a nutshell, it combines the advantages of supervised learning, where algorithms learn from data that is known and trusted to be accurately labelled, and unsupervised learning, where algorithms often deal with lots of data, but without the benefit of accurate labels. This makes it possible in situations where accurately labelled data is difficult or expensive to get, to make additional use of unlabeled data, which is often cheap and plentiful, which can bring benefits to many application areas, such as drug discovery.’
Contribution to the field
In the paper, Van Engelen and Hoos not only provide an overview of the most important and influential work that happened in the past 20 years in the field of semi-supervised learning, but also highlight fundamental concepts, approaches and insights. Hoos: ‘We wrote it for researchers and practitioners who are new to the area as well as for experts seeking to push the state of the art. Of course, we are thrilled that it was selected as one of SpringerNature’s six 2020 Research Highlights across all of Computer Science. We hope that this heightened visibility will help researchers and practitioners to benefit from the enormous potential in this fascinating and powerful set of machine learning techniques. I am also personally very pleased to see a contribution by one of my Master students recognised in this way, which highlights the quality of the students at LIACS and, more broadly, at Leiden University.’
Text: Chris Flinterman