Imbalanced classification with label noise: A systematic review and comparative analysis
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Class imbalance in datasets presents a significant challenge in machine learning, often causing traditional classification algorithms to exhibit bias toward majority classes while underrepresenting minority classes, which may be of crucial importance in various applications. This classification challenge is further exacerbated by the presence of label noise, which impedes the identification of optimal decision boundaries between classes and potentially leads to model overfitting. While extensive research has addressed class imbalance and label noise as separate phenomena, there remains a notable gap in the literature regarding their concurrent occurrence in datasets, specifically in the domain of imbalanced classification with label noise (ICLN). This review aims to bridge this gap by conducting an extensive analysis of existing methodologies addressing ICLN challenges. Our review encompasses approaches across diverse categories, including resampling techniques, ensemble methods, cost-sensitive learning, deep learning, active learning, meta-learning, and hybrid methodologies. Through rigorous empirical evaluation, we compare representative methods from each category using synthetic and real-world datasets, revealing a trade-off between minority class preservation, noise robustness, and computational efficiency. Our findings reveal that algorithm effectiveness is fundamentally dataset-dependent, with deep learning methods excelling on complex datasets while resampling approaches achieve competitive performance with lower computational cost. Statistical significance analysis validates our empirical observations, and we identify concrete future research directions for advancing ICLN methodologies.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.002 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it