Geometry and Stability of Supervised Learning Problems
Mémoli, Facundo, Brantley Vose, and Robert C. Williamson. 2024. “Geometry and Stability of Supervised Learning Problems”. ArXiv.
Search
Search results
15 results found
A Theory of Data Corruption
Corruption is notoriously widespread in data collection, and has been widely studied through the introduction of various corruption models addressing specific cases. Motivated by the existence of a zoo of models, not yet well-related or compared to each other, we worked on a framework for unifying...
Ways of The Scientist -- A lecture/seminar course
This page provides a brief summary of the course Bob will teach in Winter Semester 2024/25 at the University of Tübingen. The course is called Being a Scientist: Making Meaning by Making Science. Some content was previewed in the talk given at the IMPRS-IS bootcamp on 20 September 2024. ( slides) If...
Causal modelling without introducing counterfactuals or abstract distributions
Höltgen, Benedikt, and Robert C. Williamson. 2024. “Causal Modelling Without Introducing Counterfactuals or Abstract Distributions”. ArXiv.
An Axiomatic Approach to Loss Aggregation and an Adapted Aggregating Algorithm
Pacheco, Armando J. Cabrera, Rabanus Derr, and Robert C. Williamson. 2024. “An Axiomatic Approach to Loss Aggregation and an Adapted Aggregating Algorithm”. ArXiv.
Data Models With Two Manifestations of Imprecision
Frölich, Christian, and Robert C. Williamson. 2024. “Data Models With Two Manifestations of Imprecision”. ArXiv.
Which distribution were you sampled from? Towards a more tangible conception of data
Höltgen, Benedikt, and Robert C. Williamson. 2024. “Which Distribution Were You Sampled From? Towards a More Tangible Conception of Data”. ArXiv.
Facets of Forecast Felicity
Machine learning is about forecasting. Forecasts, however, obtain their usefulness only through their evaluation. Machine learning has traditionally focused on types of losses and their corresponding regret. Currently, the machine learning community regained interest in calibration. In this project...
Scoring Rules and Calibration for Imprecise Probabilities
Fröhlich, Christian, and Robert C. Williamson. 2024. “Scoring Rules and Calibration for Imprecise Probabilities”. ArXiv.
Thesis Projects
Below is a list of possible thesis projects within the FMLS group. These should be considered starting points only. An essential part of doing a thesis is the formulation and refinement of the problem itself. Bachelor/Master (strong focus on Math): What does predictor learn when doing ``robust...