Overview

Hyperbolic Geometry 与 Manifold Learning 将深度学习扩展到 curved spaces,利用非欧几何的特性处理 hierarchical data、scale-free networks、constrained optimization。核心洞察:hyperbolic space 的 volume 随半径指数增长 ≈ tree/hierarchy 的结构特性

核心理论

  • Poincaré Ball Model: 常用 hyperbolic space 表示,边界处距离趋于无穷
  • Lorentz Model: 另一种表示,计算更稳定
  • Riemannian Optimization: 在 manifold 上的 gradient descent(考虑 curvature)

整体趋势

  1. 从 foundational theory(Poincaré Embeddings, 2017)走向 architecture integration(Hyperbolic GNNs, Transformers)
  2. 应用场景从 knowledge graphs 扩展到 biological networks、NLP、robotics
  3. Optimization efficiency 是主要瓶颈(Riemannian operations cost 高)

技术路线

1. Hyperbolic Embeddings (Foundational)

代表工作1700-PoincareEmbeddings (🔥 Rating 5, NeurIPS 2017)

核心洞察: Hyperbolic space volume exponential growth ≈ tree depth exponential growth

关键结果

  • 5-dim Poincaré ≈ 100-dim Euclidean on WordNet
  • 20x+ dimensionality reduction
  • Link prediction SOTA

理论基础

  • Distance function: d(x,y) = arcosh(1 + 2||x-y||² / ((1-||x||²)(1-||y||²)))
  • Riemannian SGD for hyperbolic gradient descent

2. Hyperbolic Neural Networks

代表工作2400-HyperbolicNeuralNetworksSurvey (Rating 4)

核心思路: 将标准 neural layers 改写为 hyperbolic 版本

技术挑战

  • Linear operations 在 hyperbolic space 不自然(需要 tangent space 映射)
  • Attention mechanism 的 hyperbolic extension 不成熟
  • 初始化敏感

应用

  • Knowledge graph embeddings
  • Word embeddings (semantic hierarchy)
  • Hierarchical image classification

3. Hyperbolic Graph Neural Networks

代表工作2400-HyperbolicGNN (Rating 3)

核心思路: 利用 hyperbolic geometry 处理 scale-free networks

设计

  • Hyperbolic message passing
  • Hyperbolic attention
  • Manifold-aware aggregation

优势

  • Scale-free networks(citation, social)上超过 Euclidean GNN
  • 更好捕捉 hub node importance
  • 参数效率更高

局限

  • Tangent space ↔ manifold 映射 cost 高
  • 只在特定 graph types 上验证

4. Riemannian Optimization

代表工作2400-RiemannianOptimization (Rating 3)

核心思路: 在 manifold constraints 上做 gradient descent

关键 Manifolds

  • Stiefel Manifold: Orthogonal matrices({X: X^T X = I})
  • Grassmann Manifold: Subspaces(low-rank constraints)
  • Positive Definite Manifold: Covariance matrices

操作

  • Riemannian gradient(在 tangent space 计算)
  • Retraction(从 tangent space 回到 manifold)
  • Natural gradient(与 information geometry 关联)

应用

  • Orthogonal RNNs(stability)
  • Low-rank training(parameter efficiency)
  • Positive definite estimation

5. Lorentz Graph Neural Networks

代表工作2405-LorentzGNN (Rating 3)

核心思路: 将 Lorentz symmetry(相对论核心对称性)嵌入 GNN 架构

设计

  • Lorentz-equivariant message passing
  • Respect boosts/rotations 等物理变换
  • Particle physics graph structure 适配

应用

  • Particle jet classification(高能物理)
  • Physics-inspired equivariance

亮点: Equivariance 提升泛化,减少 training data 需求

6. Hyperbolic Contrastive Learning

代表工作2501-HyperbolicGraphContrastive (Rating 3)

核心思路: 在 hyperbolic space 中引入 contrastive learning,利用 hierarchical positive sampling

关键设计

  • Hierarchical positive sampling(按 hierarchy 采样 positives)
  • Geometry-aware contrastive loss
  • Curvature-aware gradient handling

应用

  • Node classification SOTA
  • Graph property prediction
  • Few-shot learning

7. Scalable Hyperbolic Knowledge Graphs

代表工作2502-ScalableHyperbolicKG (Rating 3)

核心思路: 大规模 KG 的 scalable hyperbolic embedding

工程突破

  • Mini-batch training in hyperbolic space
  • Efficient curvature learning(data-adaptive)
  • Scalable RSGD 实现

结果

  • 百万实体 KG embedding feasible
  • Link prediction improved
  • Curvature learning stable

意义: Scalability 是 hyperbolic embedding 的核心痛点,mini-batch 是重要工程突破

8. Hyperbolic Attention Networks

代表工作2503-HyperbolicAttention (Rating 3)

核心思路: Hyperbolic attention 机制捕捉 long-range dependencies

设计

  • Distance-based attention in hyperbolic space
  • Hyperbolic distance better encodes hierarchical distance
  • Geometry-aware positional encoding

理论基础

  • Attention weight = exp(-hyperbolic distance)
  • Distance ≈ hierarchy distance

应用

  • Document-level translation improved
  • Long-context language modeling better

关联: Hierarchical planning 可能受益于 hyperbolic attention

9. Accelerated Riemannian Optimization

代表工作2502-AcceleratedRiemannianOptimization (Rating 3)

核心思路: 多 geometric constraints 同时优化

挑战

  • 多 constraints 的 manifold intersection
  • Retraction 在 complex manifold 上复杂

应用

  • Orthogonal + low-rank 同时约束
  • Positive definite + Stiefel 组合

10. Fixed-Rank Matrix Manifolds

代表工作2502-RiemannianFixedRank (Rating 3)

核心思路: 研究 fixed-rank manifold 上不同 Riemannian metrics 的影响

理论贡献

  • Riemannian metrics family 分析
  • Geometry-to-algorithms pipeline
  • Metric 选择对收敛速度的影响

应用

  • Matrix completion
  • Low-rank factorization

Datasets & Benchmarks

Dataset类型SOTA特点
WordNetTaxonomyPoincaré EmbeddingsHierarchical structure
FreebaseKnowledge GraphHyperbolic KG EmbeddingsMulti-relational
Citation NetworksScale-free GraphHyperbolic GNNPower-law degree
Social NetworksScale-free GraphHyperbolic GNNHub nodes

Key Takeaways

  1. Hyperbolic ≈ Hierarchy: 双曲几何的 exponential volume growth 天然匹配 hierarchical/tree-structured data
  2. 维度效率惊人: Poincaré Embeddings 证明 5-dim hyperbolic ≈ 100-dim Euclidean
  3. Scale-free networks benefit: Hyperbolic GNN 在 power-law degree graphs 上表现优异
  4. Optimization 是瓶颈: Riemannian operations cost 高,限制了大规模应用
  5. Curvature learning 是趋势: Data-adaptive manifold curvature estimation 正在兴起

Open Problems

  1. 大规模 Scalability: Riemannian optimization 在大规模 distributed training 上的效率?
  2. Attention in Hyperbolic Space: Transformer 的 hyperbolic extension 如何设计?
  3. Curvature Learning: 如何自动选择 manifold curvature?
  4. Dynamic Hierarchies: 动态变化的 hierarchy 如何处理?
  5. Multimodal Hyperbolic: VLM/Agent 的 hyperbolic embedding 应用?
  6. Optimization Speed: 如何加速 Riemannian operations?

调研日志

  • 调研日期: 2026-04-28
  • 论文统计: 10 篇笔记(4 foundational + 6 近期 papers)
  • 覆盖范围: Hyperbolic embeddings, GNN, Contrastive Learning, Attention, Riemannian Optimization, Lorentz symmetry
  • 近一个月论文: 通过多 agent 搜索获取(部分 WebSearch 受限,使用 domain knowledge 补充)