Novel binning-based methods for model fitting and data splitting improved machine learning imbalanced data
Bibliographic record
Abstract
Abstract Machine Learning (ML) models may perform inconsistently on individual classes on nominal outputs or ranges on continuous outputs, collectively referred to here as bins. Models should be assessed through metrics that consider each bin individually, called bin metrics. Inconsistent model performance is often due to model fitting with imbalanced data. Towards improving modelling of imbalanced data, novel model fitting methods are proposed including using bin metrics as loss functions and the use of Epoch sampling. Imbalanced data also poses a challenge for appropriate data splitting. Akin split is a novel method proposed that objectively yields the most appropriate data split(s). Existing and novel model fitting methods were used to fit models, and the models were assessed by a bin metric in in two case studies. The first case study used synthetically generated datasets with different levels of noise and imbalance. On datasets with noise and greater levels of imbalance, Epoch sampling significantly improved the model performance by up to 23.6% while significantly using less resources (computation and time) by up to 57.7% compared to a standard model fitting method. The second case study used protein-genome interactions data that are often severely right-skewed. Akin split was used to split the data more appropriately than traditional methods. Model fitting methods were tried on two model configurations. The effects of the model fitting methods varied by the model configuration, but all models were significantly improved by up to 57.7% compared to the standard model fitting.
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.
How this classification was reachedexpand
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 | 0.006 | 0.004 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.009 | 0.014 |
| Research integrity | 0.001 | 0.001 |
| 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 itClassification
machine, unvalidatedMachine predicted; both teacher heads agree on what is shown here.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".