SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2409.17643 · NeurIPS · 2025

Efficient Fairness-Performance Pareto Front Computation

Shie Mannor, Mark Kozdoba, Binyamin Perets

arXiv · PDF · Open in the Atlas

Code that ran

2 of 4 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 2 failures of our instrument; 2 no recorded run), measured 2026-11-22; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2409.17643.json · badge: /api/ran/2409.17643.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
bp6725/FairPareto canonical (classifier label)
classifier basis: census
2 ran · 2 no recorded run
FunctionStatusWhere it lives
compute_group_statistics Ran
ran; our contract draft was wrong, not the code
bp6725/FairPareto/fairpareto/utils.py
code served (permissive licence) · get_code("751ccdd276284afe")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

np.random.seed(0)

# Create sample data for the function
# y: binary target labels (0 or 1)
# s: sensitive attribute values (0 or 1)

y = np.array([0, 1, 1, 0, 1, 1, 0, 0])
s = np.array([0, 0, 0, 1, 1, 1, 0, 1])

ARGS = {
    'y': y,
    's': s,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
equal_frequency_binning Ran
ran; our contract draft was wrong, not the code
bp6725/FairPareto/fairpareto/utils.py
code served (permissive licence) · get_code("ec97aaaa835e1cb1")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Set random seed for reproducibility
np.random.seed(0)

# Create sample probability values to bin
# Using values between 0 and 1 that represent probabilities
c_over_samples = np.random.uniform(0, 1, size=100)

# Number of bins to create
n_bins = 5

# Construct the arguments dictionary
ARGS = {
    "c_over_samples": c_over_samples,
    "n_bins": n_bins,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
split_to_sets Not yet run
unverified (harvested, no recorded run)
bp6725/FairPareto/fairpareto/utils.py
code served (permissive licence) · get_code("55f6ef015a968217")
How it ran

no invocation recorded.

train_calibrated_clf Not yet run
unverified (harvested, no recorded run)
bp6725/FairPareto/fairpareto/calibration.py
code served (permissive licence) · get_code("219856f7878d47fd")
How it ran

no invocation recorded.

Also present in other papers' code

Of 4 samples lifted from this paper's linked repositories, 4 carry a function identity and 0 do not. Of the 4 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 4 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

There is a well known intrinsic trade-off between the fairness of a representation and the performance of classifiers derived from the representation. In this paper we propose a new method to compute the optimal Pareto front of this trade off. In contrast to the existing methods, this approach does not require the training of complex fair representation models. Our approach is derived through three main steps: We analyze fair representations theoretically, and derive several structural properties of optimal representations. We then show that these properties enable a reduction of the computation of the Pareto Front to a compact discrete problem. Finally, we show that these compact approximating problems can be efficiently solved via off-the shelf concave-convex programming methods. In addition to representations, we show that the new methods may also be used to directly compute the Pareto front of fair classification problems. Moreover, the proposed methods may be used with any concave performance measure. This is in contrast to the existing reduction approaches, developed recently in fair classification, which rely explicitly on the structure of the non-differentiable accuracy measure, and are thus unlikely to be extendable. The approach was evaluated on several real world benchmark datasets and compares favorably to a number of recent state of the art fair representation and classification methods. result, in conjunction with an additional approximation lemma, is used in Section 4.1 to construct representations with any desired degree of approximation. Finally, based on these results in Section 4.1 we also introduce the MIFPO (Model Independent Fairness-Performance Optimization), a discrete optimisation problem that is essentially equivalent to a computation of the fairness-performance tradeoff on a discrete set. We show that in this situation MIFPO is a concave minimisation problem with linear constraints, and we solve it using the disciplined convex-concave programming framework, DCCP, Shen et al. (2016). We evaluate our approach on standard fairness benchmark datasets and compare its fairnessperformance curve to multiple state-of-the-art fair representation methods. We also compare MIFPO to the fairness-performance Pareto front of fair classifiers 1 . As expected, MIFPO effectively serves as an upper bound on almost all other algorithms in both cases. To summarise, the contributions of this paper are as follows: (a) We derive several new structural properties of optimal fair representations. (b) We use these properties to construct a model independent problem, MIFPO, which can approximate the Pareto Front of arbitrary high dimensional data distributions, but is much simpler to solve than direct representation learning for such distributions. (c) We illustrate the approach on real world fairness benchmarks. The rest of this paper is organised as follows: Section 2 discusses the literature and related work. In Section 3 we discuss the theoretical results, including factorization and the Invertibility Theorem. The MIFPO problem construction and the full Pareto Front computation algorithm are provided in Section 4. Experimental results are presented in Section 5, and we conclude the paper in Section 6. All proofs are provided in the Supplementary Material. 2 Literature and Prior Work We refer to the book Barocas et al. (2023), and surveys Mehrabi et al. (2021),Du et al. (2020), for a general overview of representations. Tradeoffs in particular where explicitly studied in Song et al. (2019), Balunović et al. (2022a), Zhao and Gordon (2022), Jovanović et al. (2023),Dehdashtian et al. (2024), among others. In this paper we use the total variation based fairness constraints, similarly to the line of work in Madras et al. (2018), Zhao and Gordon (2022), Balunović et al. (2022a), Jovanović et al. (2023). Other constraints used in the literature include entropy based constraints, Song et al. (2019), or RKHS based independence test constraints, Dehdashtian et al. (2024).

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2409.17643")
get_code_for_paper("2409.17643")
have("2409.17643")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2409.17643.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2409.17643.svg)](https://syntology.ai/paper/2409.17643)

Connect an agent — have() is free.