Pith Number

pith:CGZ3UKUO

pith:2026:CGZ3UKUOM6DYC2Y4LPNASKGWO6

not attested not anchored not stored refs resolved

Bias In, Bias Out? Finding Unbiased Subnetworks in Vanilla Models

Abdel Djalil Sad Saoud, Ekaterina Iakovleva, Enzo Tartaglione, Ivan Luiz De Moura Matos, Vito Paolo Pastore

Standard neural networks trained on biased data already contain unbiased subnetworks that can be isolated by pruning without retraining.

arxiv:2603.05582 v2 · 2026-03-05 · cs.LG · cs.CV

Open paper page JSON Open Graph Bundle Merged state Verified badge What is a Pith Number?

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{CGZ3UKUOM6DYC2Y4LPNASKGWO6}

Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge

Record completeness

1 Bitcoin timestamp

2 Internet Archive

3 Author claim open · sign in to claim

4 Citations open

5 Replications open

✓ Portable graph bundle live · download bundle · merged state

The bundle contains the canonical record plus signed events. A mirror can host it anywhere and recompute the same current state with the deterministic merge algorithm.

Claims

C1strongest claim

such subnetworks can be extracted via pruning and can operate without modification, effectively relying less on biased features and maintaining robust performance.

C2weakest assumption

That bias-free subnetworks already exist within conventionally trained models and can be reliably identified and isolated by pruning without any additional unbiased data or retraining.

C3one line summary

BISE extracts bias-free subnetworks from conventionally trained models via pruning, enabling debiased operation without retraining or additional data.

References

114 extracted · 114 resolved · 2 Pith anchors

[1] The EU artificial intelligence act, 2024 2024

[2] Does data repair lead to fair models? curating con- textually fair data to reduce model bias 2022

[3] Systematic generalisation with group in- variant predictions 2021

[4] Mind the gap: Challenges of deep learning approaches to theory of mind.Artificial Intelligence Review, 2023 2023

[5] Learning de-biased represen- tations with biased representations 2020

Receipt and verification

First computed	2026-05-18T03:09:23.072596Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

11b3ba2a8e6787816b1c5bda0928d677adc2b41feb2128b7f7fbd5c4c57266fb

Aliases

arxiv: 2603.05582 · arxiv_version: 2603.05582v2 · doi: 10.48550/arxiv.2603.05582 · pith_short_12: CGZ3UKUOM6DY · pith_short_16: CGZ3UKUOM6DYC2Y4 · pith_short_8: CGZ3UKUO

Agent API

Resolver JSON Graph JSON Events JSON Schema Signing key

Verify this Pith Number yourself

curl -sH 'Accept: application/ld+json' https://pith.science/pith/CGZ3UKUOM6DYC2Y4LPNASKGWO6 \
  | jq -c '.canonical_record' \
  | python3 -c "import sys,json,hashlib; b=json.dumps(json.loads(sys.stdin.read()), sort_keys=True, separators=(',',':'), ensure_ascii=False).encode(); print(hashlib.sha256(b).hexdigest())"
# expect: 11b3ba2a8e6787816b1c5bda0928d677adc2b41feb2128b7f7fbd5c4c57266fb

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "f762874fdb892df5881b6aa267b02aee7436d9a68466bbbede61cf3b90da3be9",
    "cross_cats_sorted": [
      "cs.CV"
    ],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.LG",
    "submitted_at": "2026-03-05T18:54:24Z",
    "title_canon_sha256": "73ad39a2088fcc1504007edb5e13d51b91ac084245ac97158a4d3a401c11e2c7"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2603.05582",
    "kind": "arxiv",
    "version": 2
  }
}