Cascade Processes for Sparse Machine Learning

Informazioni relative al progetto

SPARSE-ML

ID dell’accordo di sovvenzione: 101116395

DOI

10.3030/101116395

Data della firma CE 25 Ottobre 2023

Data di avvio 1 Dicembre 2023

Data di completamento 30 Novembre 2028

Finanziato da

• European Research Council (ERC)

Costo totale

€ 1 499 285,00

Contributo UE

€ 1 499 285,00

Coordinato da

CISPA - HELMHOLTZ-ZENTRUM FUR INFORMATIONSSICHERHEIT GGMBH
Germany

CORDIS fornisce collegamenti ai risultati finali pubblici e alle pubblicazioni dei progetti ORIZZONTE.

I link ai risultati e alle pubblicazioni dei progetti del 7° PQ, così come i link ad alcuni tipi di risultati specifici come dataset e software, sono recuperati dinamicamente da .OpenAIRE .

Pubblicazioni

Conference proceedings (31)

Robustness of Mixtures of Experts to Feature Noise

Autori: Dong Sun; Rahul Nittala; Rebekka Burkholz
Pubblicato in: Forty-third International Conference on Machine Learning, 2026, Numero 43, 2026
Editore: arXiv
DOI: 10.60882/CISPA.32771433

Pruning neural network models for gene regulatory dynamics using data and domain knowledge

Autori: Intekhab Hossain, Jonas Fischer, Rebekka Burkholz, John Quackenbush
Pubblicato in: Advances in Neural Information Processing Systems 37, 2024
Editore: Neural Information Processing Systems Foundation, Inc. (NeurIPS)
DOI: 10.60882/CISPA.31169941

Bridging Domains through Subspace-Aware Model Merging

Autori: Levy Chaves; Chao Zhou; Rebekka Burkholz; Eduardo Valle; Andra Avila
Pubblicato in: IEEE Conference on Computer Vision and Pattern Recognition, 2026, 2026
Editore: Computer Vision Foundation (CVF) / IEEE

Are GATs Out of Balance?

Autori: Nimrah Mustafa; Aleksandar Bojchevski; Rebekka Burkholz
Pubblicato in: Thirty-seventh Conference on Neural Information Processing Systems, 2023, Numero 37, 2023, ISSN 1049-5258
Editore: NeurIPS
DOI: 10.60882/CISPA.25233052

Masks, Signs, And Learning Rate Rewinding.

Autori: Advait Gadhikar; Rebekka Burkholz
Pubblicato in: The Twelfth International Conference on Learning Representations, 2024, Numero 12, 2024
Editore: ICLR
DOI: 10.60882/CISPA.26808280

Fixed Aggregation Features Can Rival GNNs

Autori: Celia Rubio-Madrigal; Rebekka Burkholz
Pubblicato in: Forty-third International Conference on Machine Learning, 2026, Numero 43, 2026
Editore: PMLR
DOI: 10.60882/CISPA.32771526

When Shift Happens - Confounding Is to Blame

Autori: Abbavaram Gowtham Reddy; Celia Rubio-Madrigal; Rebekka Burkholz; Krikamol Muandet
Pubblicato in: The Fourteenth International Conference on Learning Representations, 2026, Numero 14, 2026
Editore: International Conference on Learning Representations (ICLR)
DOI: 10.60882/CISPA.31150255

The Graphon Limit Hypothesis: Understanding Neural Network Pruning via Infinite Width Analysis

Autori: Hoang Pham; The-Anh Ta; Tom Jacobs; Rebekka Burkholz; Long Tran-Thanh
Pubblicato in: The Thirty-ninth Annual Conference on Neural Information Processing Systems, 2025, Numero 39, 2025
Editore: eural Information Processing Systems Foundation
DOI: 10.60882/CISPA.31170061

Pruning neural network models for gene regulatory dynamics using data and domain knowledge

Autori: Rebekka Burkholz, Jonas Fischer, Intekhab Hossain, John Quackenbush
Pubblicato in: Advances in Neural Information Processing Systems 37, 2025
Editore: Neural Information Processing Systems Foundation, Inc. (NeurIPS)
DOI: 10.52202/079017-3732

Spectral Graph Pruning Against Over-Squashing and Over-Smoothing

Autori: Adarsh Jamadandi, Celia Rubio-Madrigal, Rebekka Burkholz
Pubblicato in: Advances in Neural Information Processing Systems 37, 2026
Editore: Neural Information Processing Systems Foundation, Inc. (NeurIPS)
DOI: 10.60882/CISPA.31169929

SparseOpt: Addressing Normalization-induced Gradient Skew in Sparse Training

Autori: Mohammed Adnan; Rohan Jain; Tom Jacobs; Ekansh Sharma; Rahul G Krishnan; Rebekka Burkholz; Yani Ioannou
Pubblicato in: Forty-third International Conference on Machine Learning, 2026, Numero 43, 2026
Editore: OpenReview
DOI: 10.60882/CISPA.32771370

Linear GCNs Need Better Bias, Not More Expressive Power

Autori: Celia Rubio-Madrigal; Rebekka Burkhol
Pubblicato in: Workshop on Graph Foundation Models: A New Era for Graph Machine Learning (GFM) @ ICML, 2026, 2026
Editore: OpenReview
DOI: 10.60882/CISPA.32771808

HAM: A Hyperbolic Step to Regulate Implicit Bias

Autori: Tom Jacobs et al
Pubblicato in: 2026
Editore: International Con-ference on Learning Representations (ICLR)
DOI: 10.48550/ARXIV.2506.02630

Hyperbolic Aware Minimization: Implicit Bias for Sparsity

Autori: Tom Jacobs; Advait Gadhikar; Celia Rubio-Madrigal; Rebekka Burkholz
Pubblicato in: The Fourteenth International Conference on Learning Representations, 2026, Numero 41, 2026
Editore: arXiv
DOI: 10.60882/CISPA.31229218

Multi-Agent Systems are Mixtures of Experts: Who Becomes an Influencer?

Autori: Franka Bause; Jonas Niederle; Martin Pawelczyk; Rebekka Burkholz
Pubblicato in: Workshop on Compositional Learning: Safety, Interpretability, and Agents (CompLearn) @ ICML, 2026, 2026
Editore: arXiv
DOI: 10.60882/CISPA.32771649

HORST: Composing Optimizer Geometries for Sparse Transformer Training

Autori: Tom Jacobs; Rohan Jain; Rebekka Burkholz
Pubblicato in: Workshop on High-dimensional Learning Dynamics (HiLD) @ ICML, 2026, 2026
Editore: arXiv
DOI: 10.60882/CISPA.32771553

Batch normalization is sufficient for universal function approximation in CNNs

Autori: Rebekka Burkholz
Pubblicato in: The Twelfth International Conference on Learning Representations, 2024, Numero 12, 2024
Editore: International Conference on Learning Representations (ICLR)
DOI: 10.60882/CISPA.26892709

GATE: How to Keep Out Intrusive Neighbors

Autori: Nimrah Mustafa; Rebekka Burkholz
Pubblicato in: Forty-first International Conference on Machine Learning, 2024, Numero 41, 2024
Editore: PMLR
DOI: 10.60882/CISPA.27195786

Continuous Sparsification via Minimizing Movement

Autori: Hoang Pham; Tom Jacobs; Binh Nguyen; Rebekka Burkholz; Long Tran-Thanh
Pubblicato in: Workshop on High-dimensional Learning Dynamics (HiLD) @ ICML, 2026, 2026
Editore: PMLR
DOI: 10.60882/CISPA.32771496

Prune to Protect: Faster Training and Enhanced Privacy by Dynamic Data Pruning

Autori: Chinmay Joshi; Advait Gadhikar; Celia Rubio-Madrigal; Aneet Kumar Dutta; Mridula Singh; Rebekka Burkholz
Pubblicato in: Workshop on The Impact of Memorization on Trustworthy Foundation Models (MemFM) @ ICML, 2026, 2026
Editore: OpenReview
DOI: 10.60882/CISPA.32771721

Boosting for Predictive Sufficiency

Autori: Abbavaram Gowtham Reddy; Rajeev Verma; Celia Rubio-Madrigal; Krikamol Muandet; Rebekka Burkholz
Pubblicato in: The Fourteenth International Conference on Learning Representations, 2026, Numero 14, 2026
Editore: International Con-ference on Learning Representations (ICLR)
DOI: 10.60882/CISPA.31150249

Spectral Graph Pruning Against Over-Squashing and Over-Smoothing

Autori: Rebekka Burkholz, Adarsh Jamadandi, Celia Rubio-Madrigal
Pubblicato in: Advances in Neural Information Processing Systems 37, 2024
Editore: Neural Information Processing Systems Foundation, Inc. (NeurIPS)
DOI: 10.52202/079017-0331

Edge-LLMs: Edge-Device Large Language Model Competition

Autori: Shiwei Liu; Kai Han; Adriana Fernandez-Lopez; Ajay Kumar Jaiswal; Zahra Atashgahi; Boqian Wu; Edoardo Ponti; Callie Hao; Rebekka Burkholz; Olga Saukh; Lu Yin; Andreas Zinonos; Tianjin Huang; Jared Tanner; Yunhe Wang
Pubblicato in: NeurIPS 2024 Competition Track, 2024, Numero 38, 2024
Editore: NeurIPS

Dynamic Rescaling for Training GNNs

Autori: Rebekka Burkholz, Nimrah Mustafa
Pubblicato in: Advances in Neural Information Processing Systems 37, 2024
Editore: Neural Information Processing Systems Foundation, Inc. (NeurIPS)
DOI: 10.60882/CISPA.31169848

Don't Trust Stubborn Neighbors: A Security Framework for Agentic Networks

Autori: Samira Abedini; Sina Mavali; Lea Schönherr; Martin Pawelczyk; Rebekka Burkholz
Pubblicato in: Workshop on Compositional Learning: Safety, Interpretability, and Agents (CompLearn) @ ICML, 2026, 2026
Editore: PMLR
DOI: 10.60882/CISPA.32771607

Frequency-Based Hyperparameter Selection in Games

Autori: Aniket Sanyal et al.
Pubblicato in: 2026
Editore: ML Research Press (PMLR)
DOI: 10.60882/CISPA.31157677

GNNs Getting ComFy: Community and Feature Similarity Guided Rewiring

Autori: Celia Rubio-Madrigal; Adarsh Jamadandi; Rebekka Burkholz
Pubblicato in: The Thirteenth International Conference on Learning Representations, 2025, Numero 13, 2025
Editore: International Conference on Learning Representations
DOI: 10.60882/CISPA.28731815

Never Saddle for Reparameterized Steepest Descent as Mirror Flow

Autori: Tom Jacobs, Chao Zhou, and Rebekka Burkholz
Pubblicato in: The Fourteenth International Conference on Learning Representations, 2026, Numero 14, 2026
Editore: nternational Conference on Learning Representations (ICLR)
DOI: 10.60882/CISPA.31171063

Mirror, Mirror of the Flow: How Does Regular-ization Shape Implicit Bias?

Autori: Tom Jacobs, Chao Zhou, and Rebekka Burkholz
Pubblicato in: Forty-second International Conference on Machine Learning, 2025, Numero 42, 2025
Editore: Modern Language Association of America (PMLA)
DOI: 10.60882/CISPA.31169938

Mask in the Mirror: Implicit Sparsification

Autori: Tom Jacobs and Rebekka Burkholz
Pubblicato in: The Thirteenth International Conference on Learning Representations, 2025, Numero 13, 2025
Editore: arXiv
DOI: 10.60882/CISPA.31169785

Sign-In to the Lottery: Reparameterizing Sparse Training

Autori: Advait Gadhikar; Tom Jacobs; Chao Zhou; Rebekka Burkholz
Pubblicato in: The Thirty-ninth Annual Conference on Neural Information Processing Systems, 2025, Numero 39, 2025
Editore: Conference on Neural Information Processing Systems (NeurIPS)
DOI: 10.60882/CISPA.31169662

Peer reviewed articles (2)

Sex differences in gene regulation and its impact on cancer incidence

Autori: Camila M. Lopes-Ramos, Rebekka Burkholz, Marouen Ben Guebila, Viola Fanfani, Enakshi Saha, Katherine H. Shutta, Kimberly Glass, John Quackenbush, Dawn L. DeMeo
Pubblicato in: iScience, Numero 29, 2026, ISSN 2589-0042
Editore: Elsevier BV
DOI: 10.60882/CISPA.32323656

Biologically informed NeuralODEs for genome-wide regulatory dynamics

Autori: Intekhab Hossain, Viola Fanfani, Jonas Fischer, John Quackenbush, Rebekka Burkholz
Pubblicato in: Genome Biology, Numero 25, 2024, ISSN 1474-760X
Editore: Springer Science and Business Media LLC
DOI: 10.60882/CISPA.25469869