mlsanityData to 7 Oct 2026

Mapping the Multiverse of Latent Representations

arXiv 2402.01514 ↗First posted 2 Feb 2024cs.LG math.AT stat.MLcited by 5

Not on the map. Why?

Abstract (arXiv)

Echoing recent calls to counter reliability and robustness concerns in machine learning via multiverse analysis, we present PRESTO, a principled framework for mapping the multiverse of machine-learning models that rely on latent representations. Although such models enjoy widespread adoption, the variability in their embeddings remains poorly understood, resulting in unnecessary complexity and untrustworthy representations. Our framework uses persistent homology to characterize the latent spaces arising from different combinations of diverse machine-learning methods, (hyper)parameter configurations, and datasets, allowing us to measure their pairwise (dis)similarity and statistically reason about their distributions. As we demonstrate both theoretically and empirically, our pipeline preserves desirable properties of collections of latent representations, and it can be leveraged to perform sensitivity analysis, detect anomalous embeddings, or efficiently and effectively navigate hyperparameter search spaces.

Key facts. Cited by 5 papers in the map (data through 7 Oct 2026). Builds on 37 of its 86 references that are in the map. First posted on arXiv 2 Feb 2024.

Builds on

37 of 86 references.

References · 37

  1. 2017Attention Is All You Needcited by 59,932
  2. 2019Sentence-BERT: Sentence Embeddings using Siamese BERT-Networkscited by 9,049
  3. 2018UMAP: Uniform Manifold Approximation and Projection for Dimension Reductioncited by 2,841
  4. 2019Similarity of Neural Network Representations Revisitedcited by 1,695
  5. 2017Wasserstein Auto-Encoderscited by 371
  6. 2018Understanding disentangling in β-VAEcited by 332
  7. 2015Persistence Images: A Stable Vector Representation of Persistent Homologycited by 235
  8. 2018A disciplined approach to neural network hyper-parameters: Part 1 -- learning rate, batch size, momentum, and weight decaycited by 223
  9. 2022Relative representations enable zero-shot latent space communicationcited by 175
  10. 2023Similarity of Neural Network Models: A Survey of Functional and Representational Measurescited by 128
  11. 2017InfoVAE: Information Maximizing Variational Autoencoderscited by 124
  12. 2019Which Algorithmic Choices Matter at Which Batch Sizes? Insights From a Noisy Quadratic Modelcited by 116
25 more
  1. 2016Topology and Geometry of Half-Rectified Network Optimizationcited by 102
  2. 2022Reliability of CKA as a Similarity Measure in Deep Learningcited by 82
  3. 2018Neural Persistence: A Complexity Measure for Deep Neural Networks Using Algebraic Topologycited by 73
  4. 2021Representation Topology Divergence: A Method for Comparing Neural Network Representationscited by 66
  5. 2018A Topological Regularizer for Classifiers via Persistent Homologycited by 62
  6. 2022Verifying the Union of Manifolds Hypothesis for Image Datacited by 62
  7. 2013Stochastic Convergence of Persistence Landscapes and Silhouettescited by 43
  8. 2019The Shape of Data: Intrinsic Distance for Data Distributionscited by 38
  9. 2020On Implicit Regularization in β-VAEscited by 34
  10. 2021Manifold Topology Divergence: a Framework for Comparing Data Manifoldscited by 31
  11. 2023Learning Topology-Preserving Data Representationscited by 30
  12. 2019Persistent homology detects curvaturecited by 25
  13. 2020Evaluating the Disentanglement of Deep Generative Models through Manifold Topologycited by 20
  14. 2014Subsampling Methods for Persistent Homologycited by 17
  15. 2014A persistence landscapes toolbox for topological statisticscited by 16
  16. 2021Exploring the Geometry and Topology of Neural Network Loss Landscapescited by 14
  17. 2022Capturing Shape Information with Multi-Scale Topological Loss Terms for 3D Reconstructioncited by 14
  18. 2022Modeling the Machine Learning Multiversecited by 11
  19. 2022Topological Singularity Detection at Multiple Scalescited by 11
  20. 2021Exploring the Latent Space of Autoencoders with Interventional Assayscited by 10
  21. 2022Experimental Observations of the Topology of Convolutional Neural Network Activationscited by 10
  22. 2023Uncertainty Estimation of Transformers' Predictions via Topological Analysis of the Attention Matricescited by 8
  23. 2023A Normalized Bottleneck Distance on Persistence Diagrams and Homology Preservation under Dimension Reductioncited by 4
  24. 2022Manifold Characteristics That Predict Downstream Task Performancecited by 2
  25. 2023Topological Parallax: A Geometric Specification for Deep Perception Modelscited by 1

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 3; 2025: 1; 2026: 1.

5 papers.

  1. 2025No Metric to Rule Them All: Toward Principled Evaluations of Graph-Learning Datasetscited by 16
  2. 2024Diss-l-ECT: Dissecting Graph Data with Local Euler Characteristic Transformscited by 6
  3. 2024Relative Representations: Topological and Geometric Perspectivescited by 4
  4. 2026Have Graph -- Will Lift? The Case for Higher-Order Benchmarkscited by 2
  5. 2024Moments of Clarity: Streamlining Latent Spaces in Machine Learning using Moment Poolingcited by 0