README.md

metadata

license: cc-by-nc-4.0
extra_gated_heading: Acknowledge license and PhysioNet data use agreement
extra_gated_description: >-
  This dataset contains derived data from PhysioNet restricted-access datasets
  (MIMIC-CXR). By requesting access, you confirm that you have an active
  PhysioNet credentialed account and have signed the relevant data use
  agreements.
extra_gated_button_content: Request access
extra_gated_prompt: >-
  You agree to not use this dataset to conduct experiments that cause harm to
  human subjects, and you confirm compliance with the PhysioNet data use
  agreement.
extra_gated_fields:
  Full Name: text
  Affiliation: text
  Country: country
  PhysioNet Username: text
  I want to use this dataset for:
    type: select
    options:
      - Research
      - Education
      - label: Other
        value: other
  I have a valid PhysioNet credentialed account with MIMIC-CXR access: checkbox
  I agree to use this dataset for non-commercial use ONLY: checkbox
tags:
  - medical-imaging
  - chest-xray
  - embeddings
  - shortcut-detection
  - fairness
  - bias-detection
  - celeba
  - chexpert
  - mimic-cxr

ShortKit-ML — Detect and mitigate shortcuts and biases in machine learning embedding spaces. 20+ detection and mitigation methods with a unified API. Multi-attribute support tests multiple sensitive attributes simultaneously. Model Comparison mode for benchmarking multiple embedding models.

Table of Contents

• Overview
• Installation
• Quick Start
• Detection Methods
• Overall Assessment Conditions
• MCP Server
• Paper Benchmarks
• Reproducing Paper Results
• GPU Support
• Interactive Dashboard
• Testing
• Contributing
• Citation

Overview

ShortKit-ML provides a comprehensive toolkit for detecting and mitigating shortcuts (unwanted biases) in embedding spaces:

• 20+ detection methods: HBAC, Probe, Statistical, Geometric, Bias Direction PCA, Equalized Odds, Demographic Parity, Intersectional, GroupDRO, GCE, Causal Effect, SSA, SIS, CAV, VAE, Early-Epoch Clustering, and more
• 6 mitigation methods: Shortcut Masking, Background Randomization, Adversarial Debiasing, Explanation Regularization, Last Layer Retraining, Contrastive Debiasing
• 5 pluggable risk conditions: indicator_count, majority_vote, weighted_risk, multi_attribute, meta_classifier

Key Features:

• Unified ShortcutDetector API for all methods
• Interactive Gradio dashboard with real-time analysis
• PDF/HTML/Markdown reports with visualizations
• Embedding-only mode (no model access needed)
• Multi-attribute support: test race, gender, age simultaneously
• Model Comparison mode: compare multiple embedding models side-by-side

Installation

Available on PyPI at pypi.org/project/shortkit-ml.

pip install shortkit-ml

For all optional extras (dashboard, reporting, VAE, HuggingFace, MCP, etc.):

pip install "shortkit-ml[all]"

Development Install (from source)

git clone https://github.com/criticaldata/ShortKit-ML.git
cd ShortKit-ML
pip install -e ".[all]"

Or with uv:

uv venv --python 3.10
source .venv/bin/activate  # Windows: .venv\Scripts\activate
uv pip install -e ".[all]"

Optional: PDF Export Dependencies

# macOS
brew install pango gdk-pixbuf libffi
# Ubuntu/Debian
sudo apt-get install libpango-1.0-0 libpangocairo-1.0-0 libgdk-pixbuf2.0-0

HTML and Markdown reports work without these. PDF export is optional.

Quick Start

from shortcut_detect import ShortcutDetector
import numpy as np

embeddings = np.load("embeddings.npy")  # (n_samples, embedding_dim)
labels = np.load("labels.npy")          # (n_samples,)

detector = ShortcutDetector(methods=['hbac', 'probe', 'statistical', 'geometric', 'equalized_odds'])
detector.fit(embeddings, labels)

detector.generate_report("report.html", format="html")
print(detector.summary())

Embedding-Only Mode

For closed-source models or systems that only expose embeddings:

from shortcut_detect import ShortcutDetector, HuggingFaceEmbeddingSource

hf_source = HuggingFaceEmbeddingSource(model_name="sentence-transformers/all-MiniLM-L6-v2")
detector = ShortcutDetector(methods=["probe", "statistical"])
detector.fit(embeddings=None, labels=labels, group_labels=groups,
             raw_inputs=texts, embedding_source=hf_source)

See Embedding-Only Guide for CallableEmbeddingSource and caching options.

Detection Methods

Method	Key	What It Detects	Reference
HBAC	hbac	Clustering by protected attributes	-
Probe	probe	Group info recoverable from embeddings	-
Statistical	statistical	Dimensions with group differences	-
Geometric	geometric	Bias directions & prototype overlap	-
Bias Direction PCA	bias_direction_pca	Projection gap along bias direction	Bolukbasi 2016
Equalized Odds	equalized_odds	TPR/FPR disparities	Hardt 2016
Demographic Parity	demographic_parity	Prediction rate disparities	Feldman 2015
Early Epoch Clustering	early_epoch_clustering	Shortcut reliance in early reps	Yang 2023
GCE	gce	High-loss minority samples	-
Frequency	frequency	Signal in few dimensions	-
GradCAM Mask Overlap	gradcam_mask_overlap	Attention overlap with shortcut masks	-
SpRAy	spray	Spectral clustering of heatmaps	Lapuschkin 2019
CAV	cav	Concept-level sensitivity	Kim 2018
Causal Effect	causal_effect	Spurious attribute influence	-
VAE	vae	Latent disentanglement signatures	-
SSA	ssa	Semi-supervised spectral shift	arXiv:2204.02070
Generative CVAE	generative_cvae	Counterfactual embedding shifts	-
GroupDRO	groupdro	Worst-group performance gaps	Sagawa 2020
SIS	sis	Sufficient input subsets (minimal dims for prediction)	Carter 2019
Intersectional	intersectional	Intersectional fairness gaps (2+ attributes)	Buolamwini 2018

Mitigation Methods

Method	Class	Strategy	Reference
Shortcut Masking	ShortcutMasker	Zero/randomize/inpaint shortcut regions	-
Background Randomization	BackgroundRandomizer	Swap foreground across backgrounds	-
Adversarial Debiasing	AdversarialDebiasing	Remove group information adversarially	Zhang 2018
Explanation Regularization	ExplanationRegularization	Penalize attention on shortcuts (RRR)	Ross 2017
Last Layer Retraining	LastLayerRetraining	Retrain final layer balanced (DFR)	Kirichenko 2023
Contrastive Debiasing	ContrastiveDebiasing	Contrastive loss to align groups (CNC)	-

See Detection Methods Overview for per-method usage, interpretation guides, and code examples.

Overall Assessment Conditions

ShortcutDetector supports pluggable risk aggregation conditions that control how method-level results map to the final HIGH/MODERATE/LOW summary.

Condition	Best For	Description
indicator_count	General use (default)	Count of risk signals: 2+ = HIGH, 1 = MODERATE, 0 = LOW
majority_vote	Conservative screening	Consensus across methods
weighted_risk	Nuanced analysis	Evidence strength matters (probe accuracy, effect sizes, etc.)
multi_attribute	Multi-demographic	Escalates when multiple attributes flag risk
meta_classifier	Trained pipelines	Logistic regression meta-model on detector outputs (bundled model included)

detector = ShortcutDetector(
    methods=["probe", "statistical"],
    condition_name="weighted_risk",
    condition_kwargs={"high_threshold": 0.6, "moderate_threshold": 0.3},
)

Custom conditions can be registered via @register_condition("name"). See Conditions API for details.

MCP Server

ShortKit-ML ships an MCP server so AI assistants (Claude, Cursor, etc.) can call detection tools directly from chat — no Python script required.

Install the MCP extra

pip install -e ".[mcp]"

Start the server

# via entry point (after install)
shortkit-ml-mcp

# or directly
python -m shortcut_detect.mcp_server

Available tools

Tool	Description
list_methods	List all 19 detection methods with descriptions
generate_synthetic_data	Generate a synthetic shortcut dataset (linear / nonlinear / none)
run_detector	Run selected methods on embeddings — returns verdict, risk level, per-method breakdown
get_summary	Human-readable summary from a prior run_detector call
get_method_detail	Full raw result dict for a single method
compare_methods	Side-by-side comparison table + consensus vote across methods

Connect to Claude Desktop

Add the following to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS):

{
  "mcpServers": {
    "shortkit-ml": {
      "command": "python",
      "args": ["-m", "shortcut_detect.mcp_server"],
      "cwd": "/path/to/ShortKit-ML"
    }
  }
}

Paper Benchmark Datasets

Dataset 1 -- Synthetic Grid

Configure examples/paper_benchmark_config.json to control effect sizes, sample sizes, imbalance ratios, and embedding dimensionalities. A smoke profile (examples/paper_benchmark_config_smoke.json) is provided for quick sanity checks.

python -m shortcut_detect.benchmark.paper_run --config examples/paper_benchmark_config.json

Outputs CSVs, figures, and summary markdown into output/paper_benchmark/.

Dataset 2 -- CheXpert Real Data

Requires a CheXpert manifest (data/chexpert_manifest.csv) plus model-specific embedding pickles. Supported models: medclip, biomedclip, cxr-foundation.

python3 scripts/run_dataset2_benchmark.py \
  --manifest data/chexpert_manifest.csv \
  --model medclip \
  --root . \
  --artifacts-dir output/paper_benchmark/chexpert_embeddings \
  --config examples/paper_benchmark_config.json

See scripts/reproduce_paper.sh and the Dockerfile for full reproducibility.

Reproducing Paper Results

All paper results are fully reproducible with fixed seeds (seed=42). Every table and figure in the paper can be regenerated from the scripts and data in this repository.

13 benchmark methods are evaluated across all datasets: hbac, probe, statistical, geometric, frequency, bias_direction_pca, sis, demographic_parity, equalized_odds, intersectional, groupdro, gce, ssa. These span 5 paradigms: embedding-level analysis, representation geometry, fairness evaluation, explainability, and training dynamics.

Step-by-step Reproduction

Step	Command	Output	Time
1. Install	pip install -e ".[all]"	Package + deps	2 min
2. Synthetic benchmarks	python scripts/generate_all_paper_tables.py	output/paper_tables/*.tex	~10 min
3. Paper figures	python scripts/generate_paper_figures.py	output/paper_figures/*.pdf	~2 min
4. CheXpert benchmark	python scripts/run_chexpert_benchmark.py	output/paper_benchmark/chexpert_results/	~1 min
5. MIMIC-CXR setup	python scripts/setup_mimic_cxr_data.py	data/mimic_cxr/*.npy	~1 min
6. MIMIC-CXR benchmark	python scripts/run_mimic_benchmark.py	output/paper_benchmark/mimic_cxr_results/	~2 min
7. CelebA extraction	python scripts/extract_celeba_embeddings.py	data/celeba/celeba_real_*.npy	~5 min (MPS)
8. CelebA benchmark	python scripts/run_celeba_real_benchmark.py	output/paper_benchmark/celeba_real_results/	~1 min
9. Full pipeline (smoke)	./scripts/reproduce_paper.sh smoke	All synthetic outputs	~5 min
10. Full pipeline	./scripts/reproduce_paper.sh full	All synthetic outputs	~2-4 hrs

Docker (fully self-contained)

docker build -t shortcut-detect .
docker run --rm -v $(pwd)/output:/app/output shortcut-detect full

Data

Important: All embeddings and metadata are hosted here on HuggingFace. Raw CheXpert and MIMIC-CXR images and labels are not redistributed — access requires accepting the respective dataset licenses (PhysioNet for MIMIC-CXR, Stanford for CheXpert).

# Download all embeddings into data/
huggingface-cli download MITCriticalData/ShortKit-ML-data --repo-type dataset --local-dir data/

Dataset	Location	Embedding Models	Dim	Samples
Synthetic	Generated at runtime	SyntheticGenerator(seed=42)	128	Configurable
CheXpert	chexpert/	MedCLIP, ResNet-50, DenseNet-121, ViT-B/16, ViT-B/32, DINOv2, RAD-DINO, MedSigLIP	512-2048	2,000 each
MIMIC-CXR	mimic_cxr/	RAD-DINO, ViT-B/16, ViT-B/32, MedSigLIP	768-1152	~1,500 each
CelebA	celeba/	ResNet-50 (ImageNet)	2,048	10,000

Paper Tables → Scripts Mapping

Paper Table	Script	Data	Seed
Tab 3: Synthetic P/R/F1	generate_all_paper_tables.py	SyntheticGenerator	42
Tab 4: False positive rates	generate_all_paper_tables.py	SyntheticGenerator (null)	42
Tab 5: Sensitivity analysis	generate_all_paper_tables.py	SensitivitySweep	42
Tab 6: CheXpert results	run_chexpert_benchmark.py	chest_embeddings.npy	42
Tab 7: MIMIC-CXR cross-val	run_mimic_benchmark.py	mimic_cxr/*.npy	42
Tab 8: CelebA validation	run_celeba_real_benchmark.py	celeba/celeba_real_embeddings.npy	42
Tab 9: Risk conditions	generate_all_paper_tables.py	SyntheticGenerator	42
Fig 2: Convergence matrix	generate_paper_figures.py	Synthetic + CheXpert	42

See reproducibility docs for full details.

GPU Support

The library auto-selects the best available device. PyTorch components (probes, VAE, GroupDRO, adversarial debiasing, etc.) use the standard torch.device fallback:

Platform	Backend	Auto-detected
Linux/Windows with NVIDIA GPU	CUDA	Yes (torch.cuda.is_available())
macOS Apple Silicon	MPS	Partial -- pass device="mps" explicitly
CPU (any platform)	CPU	Yes (default fallback)

Note: Most detection methods (HBAC, statistical, geometric, etc.) run on CPU via NumPy/scikit-learn and do not require GPU. GPU acceleration benefits the torch-based probe, VAE, GroupDRO, and mitigation methods.

Interactive Dashboard

python app.py
# Opens at http://127.0.0.1:7860

Features: sample CheXpert data, custom CSV upload, PDF/HTML reports, model comparison tab, multi-attribute analysis.

CSV Format:

embedding_0,embedding_1,...,task_label,group_label,attr_race,attr_gender
0.123,0.456,...,1,group_a,Black,Male

See Dashboard Guide for detailed usage.

Testing

pytest tests/ -v
pytest --cov=shortcut_detect --cov-report=html

638 tests passing across all detection and mitigation methods.

Contributing

pip install -e ".[dev]"
pre-commit install

• Black for formatting (line length: 100), Ruff for linting, MyPy for types
• Pre-commit hooks run automatically; CI tests on Python 3.10, 3.11, 3.12
• New detectors must implement DetectorBase. See contributing docs and shortcut_detect/detector_template.py

Citation

@software{shortkit_ml2025,
  title={ShortKit-ML: Tools for Identifying Biases in Embedding Spaces},
  author={Sebastian Cajas, Aldo Marzullo, Sahil Kapadia, Qingpeng Kong, Filipe Santos, Alessandro Quarta, Leo Celi},
  year={2025},
  url={https://github.com/criticaldata/ShortKit-ML}
}

License

MIT License — see LICENSE

Contact

• GitHub: criticaldata/ShortKit-ML
• Issues: GitHub Issues
• Docs: criticaldata.github.io/ShortKit-ML