Improving machine learning models through enrichment of training data

Thesis

There has been a rapid development of machine learning (ML) technology over the past decade. However, trained ML models frequently encounter performance degradation when deployed in real-world applications. One major cause of this phenomenon is the inconsistencies in terms of the information captured in different data spaces. Such inconsistencies are often manifested as discrepancies between the development data space and the deployment data space, or between the applicationspecific data spac...

Expand abstract

Authors: Jin, Y

• DOI: 10.5287/ora-e2g6r1d79
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: visualization; visual analysis; machine learning; vis4ml