Table of Links Abstract and 1. Introduction Abstract and 1. Introduction Related Works


Convex Relaxation Techniques for Hyperbolic SVMs
3.1 Preliminaries
3.2 Original Formulation of the HSVM
3.3 Semidefinite Formulation
3.4 Moment-Sum-of-Squares Relaxation


Experiments
4.1 Synthetic Dataset
4.2 Real Dataset


Discussions, Acknowledgements, and References Related Works Related Works Related Works Convex Relaxation Techniques for Hyperbolic SVMs
3.1 Preliminaries
3.2 Original Formulation of the HSVM
3.3 Semidefinite Formulation
3.4 Moment-Sum-of-Squares Relaxation Convex Relaxation Techniques for Hyperbolic SVMs 3.1 Preliminaries 3.1 Preliminaries 3.2 Original Formulation of the HSVM 3.2 Original Formulation of the HSVM 3.3 Semidefinite Formulation 3.3 Semidefinite Formulation 3.4 Moment-Sum-of-Squares Relaxation 3.4 Moment-Sum-of-Squares Relaxation Experiments
4.1 Synthetic Dataset
4.2 Real Dataset Experiments Experiments 4.1 Synthetic Dataset 4.1 Synthetic Dataset 4.2 Real Dataset 4.2 Real Dataset Discussions, Acknowledgements, and References Discussions, Acknowledgements, and References Discussions, Acknowledgements, and References A. Proofs A. Proofs B. Solution Extraction in Relaxed Formulation B. Solution Extraction in Relaxed Formulation C. On Moment Sum-of-Squares Relaxation Hierarchy C. On Moment Sum-of-Squares Relaxation Hierarchy D. Platt Scaling [31] D. Platt Scaling [31] E. Detailed Experimental Results E. Detailed Experimental Results F. Robust Hyperbolic Support Vector Machine F. Robust Hyperbolic Support Vector Machine Abstract Hyperbolic spaces have increasingly been recognized for their outstanding performance in handling data with inherent hierarchical structures compared to their Euclidean counterparts. However, learning in hyperbolic spaces poses significant challenges. In particular, extending support vector machines to hyperbolic spaces is in general a constrained non-convex optimization problem. Previous and popular attempts to solve hyperbolic SVMs, primarily using projected gradient descent, are generally sensitive to hyperparameters and initializations, often leading to suboptimal solutions. In this work, by first rewriting the problem into a polynomial optimization, we apply semidefinite relaxation and sparse moment-sum-of-squares relaxation to effectively approximate the optima. From extensive empirical experiments, these methods are shown to perform better than the projected gradient descent approach. 1 Introduction On the other hand, learning and optimization on hyperbolic spaces are typically more involved than that on Euclidean spaces. Problems that are convex in Euclidean spaces become constrained non-convex problems in hyperbolic spaces. The hyperbolic Support Vector Machine (HSVM), as explored in recent studies [4, 5], exemplifies such challenges by presenting as a non-convex constrained programming problem that has been solved predominantly based on projected gradient descent. Attempts have been made to alleviate its non-convex nature through reparametrization [6] or developing a hyperbolic perceptron algorithm that converges to a separator with finetuning using adversarial samples to approximate the large-margin solution [7]. To our best knowledge, these attempts are grounded in the gradient descent dynamics, which is highly sensitive to initialization and hyperparameters and cannot certify optimality. This paper is available on arxiv under CC by-SA 4.0 Deed (Attribution-Sharealike 4.0 International) license. This paper is available on arxiv under CC by-SA 4.0 Deed (Attribution-Sharealike 4.0 International) license. available on arxiv available on arxiv Authors:
(1) Sheng Yang, John A. Paulson School of Engineering and Applied Sciences, Harvard University, Cambridge, MA (shengyang@g.harvard.edu);
(2) Peihan Liu, John A. Paulson School of Engineering and Applied Sciences, Harvard University, Cambridge, MA (peihanliu@fas.harvard.edu);
(3) Cengiz Pehlevan, John A. Paulson School of Engineering and Applied Sciences, Harvard University, Cambridge, MA, Center for Brain Science, Harvard University, Cambridge, MA, and Kempner Institute for the Study of Natural and Artificial Intelligence, Harvard University, Cambridge, MA (cpehlevan@seas.harvard.edu). Authors: Authors: (1) Sheng Yang, John A. Paulson School of Engineering and Applied Sciences, Harvard University, Cambridge, MA (shengyang@g.harvard.edu); (2) Peihan Liu, John A. Paulson School of Engineering and Applied Sciences, Harvard University, Cambridge, MA (peihanliu@fas.harvard.edu); (3) Cengiz Pehlevan, John A. Paulson School of Engineering and Applied Sciences, Harvard University, Cambridge, MA, Center for Brain Science, Harvard University, Cambridge, MA, and Kempner Institute for the Study of Natural and Artificial Intelligence, Harvard University, Cambridge, MA (cpehlevan@seas.harvard.edu).

Part of HackerNoon's growing list of open-source research papers, promoting free access to academic material.

Beyond Euclidean Space: Optimizing Hierarchical Data in Hyperbolic HSVMs

About Author

Comments

TOPICS

THIS ARTICLE WAS FEATURED IN

Related Stories

What Are Gyrocommutative Gyrogroups?

Why “Accuracy” Fails for Uplift Models (and What to Use Instead)

I Built a Poor Man’s Pixar Using Python, Gemini, and My Sense of Humor

Damian AI: A Digital Persona Built From My Work (and a Lot of Pivots)

Beyond the Hype: Why a “Trust Gap” Forced My Custom Voice AI Architecture

Gran Turismo 2026: The Great AI Showdown for Autonomous Driving

What Are Gyrocommutative Gyrogroups?

Why “Accuracy” Fails for Uplift Models (and What to Use Instead)

I Built a Poor Man’s Pixar Using Python, Gemini, and My Sense of Humor

Damian AI: A Digital Persona Built From My Work (and a Lot of Pivots)

Beyond the Hype: Why a “Trust Gap” Forced My Custom Voice AI Architecture

Gran Turismo 2026: The Great AI Showdown for Autonomous Driving

Light-Mode

Classic

Newspaper

Minty

Dark-Mode

Neon Noir

Minty

HN StartUps