Role Overview
In this thesis, you will extend and test an existing data quality framework across diverse datasets, models, and tasks, including LLMs, to evaluate its generalizability in real-world machine learning scenarios.
Responsibilities
- Extend dataset coverage using real-world and synthetic data from domains such as healthcare, finance, social media, e-commerce, and public benchmarks.
- Diversify task types, including classification, regression, and clustering.
- Broaden algorithmic scope by comparing a range of machine learning models.
- Evaluate the role of LLMs in assessing data quality across different data domains.
- Compare model responses to varying data quality degradations across model sizes and architectures.
- Add quality dimensions such as uniqueness, timeliness, accessibility, believability, and statistical measures.
- Benchmark generalizability by measuring the framework's reliability across tasks, models, and datasets.
Requirements
- Enrolled in a Master's or PhD program in Computer Science, Data Science, or a related field.
- Strong proficiency in Python and experience with machine learning libraries such as Scikit-learn, PyTorch, or TensorFlow.
- Understanding of data quality concepts and experience with data preprocessing and evaluation.
- B2 language proficiency in Dutch is required.
- Ability to work independently and conduct empirical research.