tabnetics 0.2.0

HDLSS-focused tabular learning toolkit with distribution-aware preprocessing, portfolio feature selection, and game-theoretic method aggregation.

Tabnetics

A Python toolkit for high-dimensional, low-sample-size (HDLSS) tabular classification. Tabnetics grew out of the review paper Machine learning on small size samples: A synthetic knowledge synthesis, which provided the library's initial theoretical background for small-sample learning. The library combines distribution-aware preprocessing, portfolio-based feature selection, and game-theoretic method aggregation into a single pipeline designed for settings where p >> n.

Homepage: tabnetics.org

pip install tabnetics

Licensed under Apache 2.0.

Optional integrations and benchmark backends can rely on third-party libraries with separate licenses and use terms. Those upstream terms still apply when you enable the related Tabnetics feature. Review Third-party integrations and licenses before using extras such as TabPFN, FLAML, XGBoost, LightGBM, CatBoost, MAPIE, Boruta, SHAP, or pyvinecopulib.

What's new in 0.2.0

Expanded classifier pool. The HDLSS extreme-regime pool now includes 22 classifiers across 16 families. New additions since 0.1.x:

• Bias-corrected classifiers: DBDA (Distance-Based Discriminant Analysis), GQDA (Generalized QDA with shrinkage), BC-SVM (Bias-Corrected Linear SVM) — three literature-backed estimators designed for p >> n settings where conventional LDA/QDA/SVM suffer from covariance estimation bias.
• Sparse Group Lasso Neural Network (SGLNN): a single-hidden-layer network with group-sparse input penalties, providing a nonlinear alternative in the extreme HDLSS regime.
• Confusion-Pursuit Discriminant Analysis (CPDA): an iterative confusion-pursuit classifier that targets misclassified pairs.
• Copula Discriminant Analysis (Copula-DA): a generative classifier that models class-conditional feature distributions via Gaussian copulas.
• Deep tabular backends: TabM and RealMLP are available both as lightweight numpy-only approximations and as full-fidelity PyTorch implementations via optional pytabkit integration.

Classifier oracle enhancements. The MNPO classifier oracle now supports four weighting schemes (tritrust, uniform, shapley, banzhaf) for combining multiple oracle signals when selecting the final classifier per dataset.

Benchmark expansion. The evaluation catalog has grown from 35 to 63 benchmark datasets (Val-18 + Val-19), with 49,696 total runs across 199 pipeline profiles. Seven datasets now achieve perfect classification, and 27 of 63 exceed 0.90 balanced accuracy.

TabArena general-tabular evaluation. A general profile is now available for N>>p datasets, with benchmark results against the TabArena NeurIPS 2025 leaderboard (Elo 1012.1). See TABARENA_RESULTS.md for details.

Random feature-selection baseline (Val-19). A random-subset FS baseline is now included in validation campaigns to provide a controlled null reference for feature selection method comparisons.

Pipeline fixes. The after_fs distribution-fitting stage now correctly operates on selected features with raw (untransformed) variables, fixing a semantic issue where DF could receive pre-transformed inputs.

GitHub Pages documentation site. API reference, method documentation, and results are now published at tabnetics.org via an auto-generated Jekyll site.

When to use Tabnetics

Tabnetics is built for tabular classification problems where the number of features greatly exceeds the number of samples:

• Transcriptomics — microarray and RNA-seq gene expression
• Proteomics and metabolomics — mass-spec feature matrices
• Other HDLSS settings — any structured tabular problem with p >> n

In these regimes the dominant failure modes are not model selection — they are unstable preprocessing, brittle feature selection, information leakage, and inflated validation estimates. Tabnetics addresses all four.

What Tabnetics adds to the HDLSS problem is not just another selector: it turns many unstable HDLSS choices into a multiplayer portfolio game. Feature-selection methods and classifier candidates are treated as competing players, oracle scores become the payoff structure, and the resulting MNPO equilibrium is used to select a robust portfolio under small-sample constraints.

Usage guide → · Methods & references → · Benchmark results → · TabArena results → · Announcements

Call for collaboration

We are actively looking for testers, collaborators, and co-authors to help validate Tabnetics on real-world HDLSS datasets, shape the companion article, and improve the codebase. If you work with high-dimensional tabular data — transcriptomics, proteomics, metabolomics, or similar — we would love to hear from you. See the Discussions page for ongoing conversations, or open a new thread to introduce your use case.

Citation

If you use Tabnetics in research, cite the repository for the specific version you used. The library is still under active development, and a companion paper will be published after the current testing and validation cycle is complete.

Repository URL: https://github.com/klokedm/tabnetics-public

@software{kokol_tabnetics_2026,
  author = {Kokol, Marko},
  title = {Tabnetics},
  year = {2026},
  url = {https://github.com/klokedm/tabnetics-public}
}

Benchmark results

Tabnetics has been evaluated on 35 HDLSS benchmark datasets (50–2,600 samples, 500–100,000 features, 2–14 classes) drawn from OpenML, GEO, Scikit-feature, and UCSC Xena/TCGA. Across 2,800+ runs with 9 random seeds per dataset, the pipeline achieves a mean balanced accuracy of 0.80, with 12 of 35 datasets above 0.90 and 3 reaching perfect classification. The MNPO portfolio consistently outperforms single-method baselines, and distribution-aware preprocessing contributes a small but consistent positive effect. Detailed per-dataset results, statistical comparisons, and dataset provenance are available in RESULTS.md. A peer-reviewed article with full methodology and ablation studies is in preparation.

General tabular data (TabArena). To provide a transparent reference on non-HDLSS data, Tabnetics was evaluated on the TabArena benchmark, a NeurIPS 2025 suite of 38 general tabular classification datasets. Using the general profile with XGBoost and LightGBM enabled, and scoring the merged local artifact with the same Elo-based leaderboard machinery used by TabArena, the current informational snapshot gives tabnetics (general) an overall Elo 1012.1. A dedicated general_tabular profile is now available for N>>p datasets — it uses a tree-weighted classifier pool, skips HDLSS-specific machinery (screening, folding, CDF transforms), and adapts feature selection by the sample-to-feature ratio. Full results and caveats are in TABARENA_RESULTS.md; profile details are in USING.md.

Key ideas

1. Distribution-aware preprocessing. Each feature is fitted to a parametric family (from 20+ candidates) using goodness-of-fit testing, bootstrap calibration, and L-moment prescreening. CDF-based transforms replace ad-hoc normalization.

2. Portfolio feature selection. Thirty feature-selection methods — stability selectors, copula knockoffs, tree-based importance, mutual-information filters, IPSS, HSIC-Lasso, and more — are run together. A game-theoretic oracle (MNPO — Nash Multi-Portfolio Optimization) aggregates their outputs into a single robust HDLSS feature portfolio. MNPO builds pairwise preference matrices from multiple oracles (performance, stability, complexity, etc.) and solves for a Nash equilibrium via KL-regularized mirror descent. The multiplayer game framing draws conceptual inspiration from Wu et al.'s Multiplayer Nash Preference Optimization, though the HDLSS adaptation is a distinct contribution with different players, oracles, and data regime (see BACKGROUND.md for details).

3. Regime-aware classification. An MNPO-based classifier oracle picks from regime-appropriate pools. The HDLSS extreme pool includes 22 classifiers spanning linear/GLM, LDA, SVM, PLS-DA, NSC, naive Bayes, random projection, distance-based DA, QDA, neural network, kernel approximation, subspace/manifold, robust distance, copula/generative, shrinkage/confusion-pursuit, and deep tabular families. The deep tabular backends (TabM, RealMLP) are available as both lightweight numpy-only approximations and as full-fidelity PyTorch implementations via optional pytabkit integration. Moderate-regime pools add RF, XGBoost, CatBoost, and TabPFN. All optional backends (CatBoost, LightGBM, TabPFN, pytabkit) degrade gracefully when not installed.

4. Strict validation. All learned preprocessing and selection is train-only. The HuggingFace bundle is the authoritative reproducibility mirror of the public upstream datasets used for validation. Synthetic fallback is not allowed for evidence-bearing runs.

Quick start

from tabnetics.pipeline import DistributionFeatureSelectionPipeline, DFFSConfig

config = DFFSConfig(random_seed=42)
pipeline = DistributionFeatureSelectionPipeline(config)

result = pipeline.run(X, y, dataset_name="my_dataset", seed=42)

print(f"Accuracy: {result.accuracy:.3f}")
print(f"Selected features: {result.selected_features}")

Operational defaults

The packaged runtime currently follows the promoted post-review workflow:

• df_stage_position="after_fs" is the default, so distribution fitting runs on the feature space that actually survives selection.
• Evidence-bearing benchmark and validation runs treat the HuggingFace bundle as the authoritative reproducibility mirror of the public upstream datasets and default to dataset_integrity_policy="error".
• Conformal prediction is opt-in and should be interpreted as an uncertainty layer (coverage, prediction-set size, singleton rate), not as a balanced-accuracy optimizer.
• multiomics_adapter="split_halves" is a benchmark-time shortcut; real multi-omics studies should use explicit blocks with tabnetics.multiomics.

Command line workflows

Editable installs expose installed wrappers, and every wrapper has the same packaged python -m ... equivalent:

tabnetics-benchmark --datasets leukemia_golub --seeds 11 23 37
tabnetics-validation-plan --plan-kind validation17 --num-pods 4
tabnetics-validation-suite --dataset-sets fs_easy --seeds 11 23 37

The corresponding module entrypoints are:

• python -m tabnetics.benchmarks.cli
• python -m tabnetics.validation.generate_plan
• python -m tabnetics.validation.core.shard_runner
• python -m tabnetics.validation.suite

TabArena benchmark (general tabular data evaluation against leaderboard-best methods):

python -m experiments.benchmarking.tabarena_benchmark \
    --dataset-sets all --profile general --seeds 42 \
    --max-workers 12 --max-train-samples 50000 --task-timeout-sec 3600

TabArena results page regeneration (official-style leaderboard scoring for the local artifact):

python -m scripts.analysis.generate_tabarena_results \
    --results-csv run_artifacts/tabarena_general_archml/tabarena_results.csv \
    run_artifacts/tabarena_general_archml_rerun6/tabarena_results.csv \
    --output-dir run_artifacts/tabarena_general_archml \
    --write-markdown core/TABARENA_RESULTS.md

See TABARENA_RESULTS.md for results and interpretation.

Selected literature anchors

The full methods table lives in BACKGROUND.md. For a quick orientation, these are the main papers behind the current public positioning:

• Kokol. Machine learning on small size samples: A synthetic knowledge synthesis — the original HDLSS review context behind the library.
• Freund & Schapire. Adaptive game playing using multiplicative weights — the mirror-descent / multiplicative-weights foundation used by the MNPO solver.
• Wu et al. Multiplayer Nash Preference Optimization — conceptual inspiration for the multiplayer Nash framing, not the solver implementation.
• Singh et al. DIABLO and Rohart et al. MINT — the reference points for explicit multi-omics integration.
• Taquet et al. MAPIE: an open-source library for distribution-free uncertainty quantification — the conformal/UQ reference behind the classifier-side uncertainty outputs.

Package structure

Subpackage	Purpose
tabnetics.core	MNPO game-theoretic primitives, sklearn compatibility layer, runtime configuration
tabnetics.distribution	Univariate distribution fitting (20+ families), bootstrap GOF, CDF-based transforms
tabnetics.feature_selection	30 selection methods, MNPO portfolio aggregation, copula knockoffs, stability selectors
tabnetics.classification	Regime-aware classifier pools, MNPO classifier oracle, PLS-DA, conformal helpers
tabnetics.pipeline	End-to-end DF+FS+classification pipeline with leakage prevention
tabnetics.datasets	Dataset registry, HuggingFace/OpenML loaders, meta-feature extraction
tabnetics.domains	Domain adapters (bioinformatics prefilters, face-domain projection)
tabnetics.multiomics	Multi-block PLS-DA (DIABLO-style) and MINT batch-correction integration
tabnetics.benchmarks	Benchmark runner, method-set profiles, SOTA comparison, gaming detection
tabnetics.validation	Validation campaign planner, shard execution, promotion gates

Feature selection methods

The FeatureSelector supports 30 methods out of the box, including:

Category	Methods
Stability selectors	Lasso stability, subspace stability, decorrelated stability, cluster stability, TIGRESS
Wrapper methods	RFECV (SVM, RF, LR), Boruta
Filter methods	ANOVA F-test, mutual information, mRMR, JMI, CMIM, FCBF, Wilcoxon AUC
Tree-based	GBDT importance, TreeSHAP, random forest
Knockoff methods	Copula knockoff (D-vine, FDR-controlled via e-values), derandomized knockoffs
Embedded	OA-Elastic Net, Joint AUC+L1, HSIC-Lasso
Other	IPSS, k-TSP, OVA/ECOC wrappers, Rashomon importance

Methods are aggregated via MNPO with configurable oracle presets (minimal, perf_only, perf_complexity, full, etc.).

See BACKGROUND.md for the full list of implemented papers, USING.md for detailed usage, and RESULTS.md for benchmark results.

Installation

Core dependencies (numpy, pandas, scipy, scikit-learn):

pip install tabnetics

With optional feature-selection extras (boruta, copula support, conformal prediction):

pip install tabnetics[feature-selection-optional]

With full benchmark support (FLAML, LightGBM, XGBoost, TabPFN, etc.):

pip install tabnetics[benchmarks]

Requirements

• Python >= 3.11
• numpy, pandas, scipy, scikit-learn (core)
• See pyproject.toml for optional dependency groups

Development

git clone https://github.com/klokedm/tabnetics-public.git
cd tabnetics-public
pip install -e ".[dev]"
pytest

License

Apache 2.0 — see LICENSE.

---

This documentation is auto-generated from internal notes and sources with the support of rule-based transformations and generative AI. Errors are possible — please report any issues via Discussions.