RINO: Renormalization Group Distillation with No Labels

Abstract

Supervised machine learning in high-energy physics (HEP) requires training on labeled simulations of particle collisions, which are known to be unfaithful to the underlying physics. This leads to significant inaccuracies and biases in the transfer of trained models to real data, limiting the power of deep learning methods. Self-supervised learning (SSL) is a promising alternative for pretraining directly on real, unlabeled data. However, existing techniques largely operate with inductive biases inspired by domains such as images and natural language. We propose RINO (Renormalization Group DIstillation with NO Labels), which, for the first time, exploits the fundamentally multi-scale renormalization group structure of high-energy jet formation as a physically grounded source of SSL views. We define teacher-student representations across energy scales using different levels of hierarchical clustering, aligned via DINO-style self-distillation. To mirror the background-dominated nature of real collider data, RINO pretrains exclusively on jets from quantum chromodynamics (QCD) processes and is evaluated on out-of-distribution (OOD) transfer to rare signal classes. In our controlled data-efficiency study, RINO matches fully supervised training on JetNet after finetuning on only 1% of the labeled data. With the full labeled dataset, RINO improves OOD classification accuracy on top-quark jets by 5.9% and, without retraining, improves AUC on the unseen Higgs boson decay channel H → 4q by 4.1% over the strongest SSL baseline. Scale-structure diagnostics reveal that DINO-style self-distillation permits our backbone to retain scale-dependent structure. Ablation studies show that this accounts for an 8.6% OOD accuracy gap with respect to alternative SSL paradigms such as SimCLR. Code is available at https://github.com/zichunhao/RINO.

The Renormalization Group as a Stochastic Exploration Process
Denis Bernard
HAL11 Aug 2026PreprintLicence not statedFull text3 citedMathematical physics General physicsHAL
Multiloop functional renormalization group from single bosons
Kilian Fraboulet, Aiman Al-Eryani, Sarah Heinzelmann +2
Physical Review Research6 Oct 2026Peer-reviewedCC BY 4.0High-energy & particle physicsCrossrefAPS
Fixed points of the CKM matrix renormalization group running to all orders in perturbation theory
Brian P Dolan
arXiv28 Aug 2026PreprintCC BY 4.0Full textHigh-energy & particle physicsarXiv
Sample-based Quantum Diagonalization via Density Matrix Renormalization Group
Yuto Sugimoto, Tomonori Shirakawa, Yukio Kawashima +2
arXiv4 Nov 2026PreprintarXiv licenceFull textQuantum physics & quantum informationarXiv
Evidence for an Obstruction to Perturbative Renormalization Group in Stochastic Turbulence
L. Ts. Adzhemyan, D. A. Davletbaeva, D. A. Evdokimov +1
arXiv1 Oct 2026PreprintarXiv licenceFull textStatistical & nonlinear physicsarXiv

Cite

BibTeX RIS

BibTeX

@misc{hao2026rino,
  title = {{RINO: Renormalization Group Distillation with No Labels}},
  author = {Zichun Hao and Raghav Kansal and Abhijith Gandrakota and Chang Sun and Haoyang Li and Zihan Zhao and Ngadiuba Jennifer and Javier Duarte and Maria Spiropulu},
  howpublished = {arXiv},
  year = {2026},
  month = oct,
  doi = {10.48550/arxiv.2610.15919},
  eprint = {2610.15919},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2610.15919},
  note = {CC BY 4.0},
}

RIS

TY  - UNPB
TI  - RINO: Renormalization Group Distillation with No Labels
AU  - Zichun Hao
AU  - Raghav Kansal
AU  - Abhijith Gandrakota
AU  - Chang Sun
AU  - Haoyang Li
AU  - Zihan Zhao
AU  - Ngadiuba Jennifer
AU  - Javier Duarte
AU  - Maria Spiropulu
PY  - 2026
DA  - 2026/10/11
PB  - arXiv
DO  - 10.48550/arxiv.2610.15919
UR  - https://arxiv.org/abs/2610.15919
AB  - Supervised machine learning in high-energy physics (HEP) requires training on labeled simulations of particle collisions, which are known to be unfaithful to the underlying physics. This leads to significant inaccuracies and biases in the transfer of trained models to real data, limiting the power of deep learning methods. Self-supervised learning (SSL) is a promising alternative for pretraining directly on real, unlabeled data. However, existing techniques largely operate with inductive biases inspired by domains such as images and natural language. We propose RINO (Renormalization Group DIstillation with NO Labels), which, for the first time, exploits the fundamentally multi-scale renormalization group structure of high-energy jet formation as a physically grounded source of SSL views. We define teacher-student representations across energy scales using different levels of hierarchical clustering, aligned via DINO-style self-distillation. To mirror the background-dominated nature of real collider data, RINO pretrains exclusively on jets from quantum chromodynamics (QCD) processes and is evaluated on out-of-distribution (OOD) transfer to rare signal classes. In our controlled data-efficiency study, RINO matches fully supervised training on JetNet after finetuning on only 1% of the labeled data. With the full labeled dataset, RINO improves OOD classification accuracy on top-quark jets by 5.9% and, without retraining, improves AUC on the unseen Higgs boson decay channel H → 4q by 4.1% over the strongest SSL baseline. Scale-structure diagnostics reveal that DINO-style self-distillation permits our backbone to retain scale-dependent structure. Ablation studies show that this accounts for an 8.6% OOD accuracy gap with respect to alternative SSL paradigms such as SimCLR. Code is available at https://github.com/zichunhao/RINO.
ER  -

Abstract

Related

Cite