CRISP: Scalable Importance-Stratified Coresets for Imbalanced Tabular Learning

You already need task-specific training and want to reduce its memory or compute cost.

What you would change

Change only the training technique on a small controlled run after checking prerequisites.

Change only the training technique on a small controlled run after checking prerequisites.

Compare against

Your current training recipe with the same model, data split, and evaluation.

1. Locate the paper's implementation and check its license, model support, data needs, and hardware requirements. Stop here if you cannot obtain the required artifacts.
2. Save a small set of representative product tasks and expected outcomes. Keep evaluation tasks out of any training or tuning, and choose quality and cost limits before running.
3. Run the baseline and the proposed change on identical inputs. Keep other settings fixed and repeat paired runs to expose variation.
4. Save per-task outputs, failures, timings, and all usage costs. Review regressions before expanding the test.

What to measure

Held-out quality, peak memory, training time, and full compute cost to the target quality.

When to keep it

Keep the change only if it meets your preselected quality and latency limits and improves the metric you care about after setup and operating costs. Otherwise retain the baseline. This is an exploration plan, not a validated recommendation.

recommendation

test

research area

efficient training

technical change

Importance-stratified coreset pruning for GBDT training

published

2026-09-22

The method fits constrained infrastructure (linear-time, no large compute) and shows strong gains on fraud-style tabular data, but the mixed public-benchmark results and dependence on a proxy model require validation on the reader's own dataset before production adoption.

1. claim 01Paperraft inference
   Read the brief interpretation

   CRISP replaces full-data gradient-boosted tree training on imbalanced tabular datasets with training on a small importance-stratified coreset of the majority class, using quantile-stratified budget allocation from a proxy-model score and inverse-propensity sample weights. It costs one proxy-model scoring pass over the data (linear time), added pipeline complexity for stratification and weighting, and a small accuracy loss (about 0.3% AP in the reported production case). It can fail when the proxy score is uninformative or the class distribution shifts, as suggested by the mixed results on Sparkov at lower reduction rates, where the method only led at extreme (99.2%+) pruning.

   This is a Paperraft inference. It is not pinned to a verified PDF passage.

   Inspect the source
2. claim 02Paperraft inference

   At 95% negative-class reduction on a production fraud dataset, CRISP trains on ~1.70M of 25M rows (93.2% fewer training rows) while retaining 99.7% of full-data Average Precision, and achieves the highest mean AP among compared methods at 90%-99.4% majority reduction on CriteoPrivateAds.

   This is a Paperraft inference. It is not pinned to a verified PDF passage.

   Inspect the source