Studying meta-features at an instance-level

• Grant number: 25/10304-4
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: July 01, 2025
• End date: June 30, 2026
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Ana Carolina Lorena
• Grantee: Thiago Galante Pereira
• Host Institution: Divisão de Ciência da Computação (IEC). Instituto Tecnológico de Aeronáutica (ITA). Ministério da Defesa (Brasil). São José dos Campos , SP, Brazil
• Associated research grant: 21/06870-3 - Beyond algorithm selection: meta-learning for data and algorithm analysis and understanding, AP.JP2

Abstract

Meta-learning (MtL) is usually focused on analyzing a collection of datasets and how their characteristics influence Machine Learning (ML) classification performance. However, using this framework at a more fine-grained instance level is also possible, where characteristics from each observation (instance) in a dataset are related to algorithmic performance. Herewith, one can explore for which instances in a dataset the algorithms struggle more and why. Another possibility isdynamically choosing the predictors for each instance based on their profiles. For such, extracting meaningful meta-features from the dataset observations is necessary. This research will study the effectiveness of meta-features at an instance level in characterizing different data types. This will allow us to broaden our understanding of how meta-features behave when describing different data types. (AU)