Pith Number

pith:23FV3LEW

pith:2026:23FV3LEWRLL65SR6MFXZX5WKYJ

not attested not anchored not stored refs resolved

Breaking Global Self-Attention Bottlenecks in Transformer-based Spiking Neural Networks with Local Structure-Aware Self-Attention

Hangming Zhang, Lingdong Li, Qiang Yu

Local dilated-window self-attention and spiking response pooling let transformer-based spiking networks preserve regional features while cutting quadratic redundancy.

arxiv:2605.13887 v1 · 2026-05-12 · cs.NE · cs.AI

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

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{23FV3LEWRLL65SR6MFXZX5WKYJ}

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

On the more challenging static dataset Tiny-ImageNet and neuromorphic dataset N-CALTECH101, LSFormer substantially outperforms state-of-the-art baselines by 4.3% and 8.6% in top-1 classification accuracy, respectively.

C2weakest assumption

That the reported accuracy improvements arise from the local structure-aware mechanism and spiking response pooling rather than from unstated hyper-parameter tuning, longer training, or dataset-specific engineering choices not controlled in the baselines.

C3one line summary

LSFormer uses local structure-aware spiking self-attention and spiking response pooling to cut global attention bottlenecks, delivering 4.3% and 8.6% accuracy gains on Tiny-ImageNet and N-CALTECH101 over prior transformer-based SNNs.

References

73 extracted · 73 resolved · 1 Pith anchors

[1] S. B. Furber, F. Galluppi, S. Temple, and L. A. Plana, “The spinnaker project,”Proceedings of the IEEE, vol. 102, no. 5, pp. 652–665, 2014 2014

[2] Trainable spiking-yolo for low-latency and high-performance object detection, 2024

[3] Event stream learning using spatio-temporal event surface, 2022

[4] Event-driven spiking neural network based on membrane potential modulation for remote sensing image classification, 2023

[5] Spiking neural network-based multi- task autonomous learning for mobile robots, 2021

Formal links

2 machine-checked theorem links

Receipt and verification

First computed	2026-05-17T23:39:19.119099Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

d6cb5dac968ad7eeca3e616f9bf6cac2496204447e9e398b754355ae26cd3302

Aliases

arxiv: 2605.13887 · arxiv_version: 2605.13887v1 · doi: 10.48550/arxiv.2605.13887 · pith_short_12: 23FV3LEWRLL6 · pith_short_16: 23FV3LEWRLL65SR6 · pith_short_8: 23FV3LEW

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/23FV3LEWRLL65SR6MFXZX5WKYJ \
  | 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: d6cb5dac968ad7eeca3e616f9bf6cac2496204447e9e398b754355ae26cd3302

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "16f34e93d0167b2c5c603890f9d123b9b2749215f01940c5d0fdcf341e129b09",
    "cross_cats_sorted": [
      "cs.AI"
    ],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.NE",
    "submitted_at": "2026-05-12T02:08:08Z",
    "title_canon_sha256": "2aa866a479987b95e0323ff3bc7f78c43a384ea8c9ef779527f1408d3870a06c"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2605.13887",
    "kind": "arxiv",
    "version": 1
  }
}