Lune

NeurIPS2024Top-tier venue

Latent Functional Maps: a spectral framework for representation alignment

Marco Fumero, Marco Pegoraro, Valentino Maiorca, Francesco Locatello, Emanuele Rodolà

Abstract

Neural models learn data representations that lie on low-dimensional manifolds, yet modeling the relation between these representational spaces is an ongoing challenge. By integrating spectral geometry principles into neural modeling, we show that this problem can be better addressed in the functional domain, mitigating complexity, while enhancing interpretability and performances on downstream tasks. To this end, we introduce a multi-purpose framework to the representation learning community, which allows to: (i) compare different spaces in an interpretable way and measure their intrinsic similarity; (ii) find correspondences between them, both in unsupervised and weakly supervised settings, and (iii) to effectively transfer representations between distinct spaces. We validate our framework on various applications, ranging from stitching to retrieval tasks, and on multiple modalities, demonstrating that Latent Functional Maps can serve as a swiss-army knife for representation alignment.

Ask Lune

Know this paper in minutes, not an afternoon.

Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.

12 top-tier papers have cited it since. Know what they found before your reviewers do.

Questions to start from

Lune can draw on

the full paper · the top-tier papers that cite it · related work

Get the answer

Free to start. No credit card required.

Lineage

  • Connecting Neural Models Latent Geometries with Relative Geodesic Representations cites Representation Topology Divergence: A Method for Comparing Neural Network Representations
  • Connecting Neural Models Latent Geometries with Relative Geodesic Representations cites An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
  • Connecting Neural Models Latent Geometries with Relative Geodesic Representations cites When Does Closeness in Distribution Imply Representational Similarity? An Identifiability Perspective
  • Connecting Neural Models Latent Geometries with Relative Geodesic Representations cites On Linear Identifiability of Learned Representations
  • Connecting Neural Models Latent Geometries with Relative Geodesic Representations cites Revisiting Model Stitching to Compare Neural Representations
  • When Does Closeness in Distribution Imply Representational Similarity? An Identifiability Perspective cites On Linear Identifiability of Learned Representations
  • When Does Closeness in Distribution Imply Representational Similarity? An Identifiability Perspective cites Revisiting Model Stitching to Compare Neural Representations
  • When Does Closeness in Distribution Imply Representational Similarity? An Identifiability Perspective cites Generalized Shape Metrics on Neural Representations
  • Revisiting Model Stitching to Compare Neural Representations cites An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
  • λ-Orthogonality Regularization for Compatible Representation Learning cites An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
  • Generalized Shape Metrics on Neural Representations cites On Linear Identifiability of Learned Representations
Builds onCited byLeft to right by year; the more cited, the higher.

Cited by top-tier papers12

  1. Navigating the Latent Space Dynamics of Neural ModelsMarco Fumero, Luca Moschella, Emanuele Rodolà, Francesco LocatelloICLR 2026 · 10 citations
  2. λ-Orthogonality Regularization for Compatible Representation LearningSimone Ricci, Niccolò Biondi, Federico Pernici, Ioannis Patras et al.NeurIPS 2025 · 8 citations
  3. When Does Closeness in Distribution Imply Representational Similarity? An Identifiability PerspectiveBeatrix M. G. Nielsen, Emanuele Marconato, Andrea Dittadi, Luigi GreseleNeurIPS 2025 · 7 citations
  4. DiffuMatch: Category-Agnostic Spectral Diffusion Priors for Robust Non-Rigid Shape MatchingEmery Pierson, Lei Li, Angela Dai, Maks OvsjanikovICCV 2025 · 5 citations
  5. Connecting Neural Models Latent Geometries with Relative Geodesic RepresentationsHanlin Yu, Berfin Inal, Georgios Arvanitidis, Søren Hauberg et al.NeurIPS 2025 · 5 citations

Builds on16

  1. An Image is Worth 16x16 Words: Transformers for Image Recognition at ScaleAlexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn et al.ICLR 2021 · 21,477 citations
  2. Revisiting Model Stitching to Compare Neural RepresentationsYamini Bansal, Preetum Nakkiran, Boaz BarakNeurIPS 2021 · 253 citations
  3. Generalized Shape Metrics on Neural RepresentationsAlex H. Williams, Erin Kunz, Simon Kornblith, Scott W. LindermanNeurIPS 2021 · 182 citations
  4. On Linear Identifiability of Learned RepresentationsGeoffrey Roeder, Luke Metz, Durk KingmaICML 2021 · 107 citations
  5. Representation Topology Divergence: A Method for Comparing Neural Network RepresentationsSerguei Barannikov, Ilya Trofimov, Nikita Balabin, Evgeny BurnaevICML 2022 · 69 citations

Related papers

  1. Functional Alignment Can Mislead: Examining Model StitchingDamian Smith, Harvey Mannering, Antonia MarcuICML 2025
  2. NAM: Neural Adjoint Maps for refining shape correspondencesGiulio Viganò, Maks Ovsjanikov, Simone MelziSIGGRAPH 2025 · 7 citations
  3. Spatially and Spectrally Consistent Deep Functional MapsMingze Sun, Shiwei Mao, Puhua Jiang, Maks Ovsjanikov et al.ICCV 2023 · 37 citations
  4. An Elastic Basis for Spectral Shape CorrespondenceFlorine Hartwig, Josua Sassen, Omri Azencot, Martin Rumpf et al.SIGGRAPH 2023 · 21 citations
  5. Geometric Embedding Alignment via Curvature Matching in Transfer LearningSung Moon Ko, Jaewan Lee, Sumin Lee, Soorin Yim et al.ICML 2026
Dusk over the sea between two cliffs drawn in fine vertical lines