A Theory of Learning with Corrupted Labels
Authors: Brendan van Rooyen, Robert C. Williamson
JMLR 2017 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Theoretical | Here we begin to develop an abstract framework for tackling these problems. We present a generic method for learning from a fixed, known, reconstructible corruption, along with an analyses of its statistical properties. We demonstrate the utility of our framework via concrete novel results in solving supervised learning problems wherein the labels are corrupted, such as learning with noisy labels, semi-supervised learning and learning with partial labels. The concrete contributions of this paper are: A general method for learning from corrupted labels based on a generalization of the method of unbiased estimators... Upper and lower bounds on the risk of learning from combinations of corrupted labels... Demonstration of the computational feasibility of our approach via the preservation of convexity... |
| Researcher Affiliation | Academia | Brendan van Rooyen EMAIL Robert C. Williamson EMAIL The Australian National University and Data61 Canberra ACT 2601, Australia |
| Pseudocode | No | The paper describes mathematical methods and algorithms using prose and mathematical notation (e.g., AERM( S) = arg min a A ℓR( S, a)), but does not contain any structured pseudocode blocks or algorithms labeled as such. |
| Open Source Code | No | The paper does not contain any explicit statements about releasing source code, nor does it provide links to any code repositories or supplementary materials for code. |
| Open Datasets | No | The paper primarily presents a theoretical framework and uses general problem types like 'Learning with Label Noise', 'Semi-Supervised Learning', and 'Learning with partial labels' as examples. It does not describe experiments performed on specific publicly available datasets. |
| Dataset Splits | No | The paper is theoretical and does not describe empirical experiments involving datasets. Therefore, there are no mentions of dataset splits (training, validation, test) or cross-validation setups. |
| Hardware Specification | No | The paper is theoretical and focuses on mathematical framework and proofs. It does not describe any computational experiments or mention specific hardware used for any part of the research. |
| Software Dependencies | No | The paper is theoretical and does not describe any computational experiments that would require specific software or library versions. Thus, no software dependencies are mentioned. |
| Experiment Setup | No | The paper is theoretical and develops a conceptual framework and mathematical bounds. It does not include any experimental results from a practical implementation, and therefore, no experimental setup details like hyperparameters, training configurations, or system-level settings are provided. |