AI Paper Atlas

Research map / Benchmarks / COMPAS

COMPAS leaderboard

14 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on COMPAS. Best accuracy: 0.881 (FairPC, AAAI 2021).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

COMPAS · accuracy

14 papers, each with its best reported number for its own method; higher is better.

#accuracy ↑MethodPaperTopic
10.881FairPCGroup Fairness by Probabilistic Modeling with Latent Fair Decisions
AAAI 2021 · Table: Comparison on real worl
Variational and generative sampling
20.879δ-fair
Naive Bayes
Learning Fair Naive Bayes Classifiers by Discovering and Eliminating Discrimination Patterns
AAAI 2020 · Table: accuracy
Fairness
30.8OursFed
3-layer NN
OursFed: Provable Group Fairness-Aware Federated Learning Against Distrust and Fragility
AAAI 2026 · Table 1
Personalized and heterogeneous FL
40.76RF+Ours
RF
Leveraging Labeled and Unlabeled Data for Consistent Fair Binary Classification
NeurIPS 2019 · Table 1
Fairness
50.724Generalized orthogonalizationGeneralizing Orthogonalization for Models with Non-Linearities
ICML 2024 · Table: Test set prediction per
Prediction and decision losses
60.6945GCMLearning Monotonic Probabilities with a Generative Cost Model
ICML 2025 · Table 2
Variational and generative sampling
70.682FairBatch
Logistic regression
FairBatch: Batch Selection for Model Fairness
ICLR 2021 · ED results table
Fairness
80.68Doubly-RobustDoubly Robust Counterfactual Classification
NeurIPS 2022 · Table: AUC and classification
Molecules and proteins
90.66Born-Again Tree
Random Forest
Born-Again Tree Ensembles
ICML 2020 · Table 4
Prediction and decision losses
100.657Ours+FC
logistic regression
Improving Fair Training under Correlation Shifts
ICML 2023 · Table 2
Fairness
4 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.