返回
Adaptive indefinite kernels in hyperbolic spaces
DOI:10.1016/j.neunet.2024.106803.png)
摘要
En 中文
Learning embeddings in hyperbolic space has gained increasing interest in the community, due to its property of negative curvature, as a way of encoding data hierarchy. Recent works investigate the improvement of the representation power of hyperbolic embeddings through kernelization. However, existing developments focus on defining positive definite (pd) kernels, which may affect the intriguing property of hyperbolic spaces. This is due to the structures of hyperbolic spaces being modeled in indefinite spaces (e.g., Krein space). This paper addresses this issue by developing adaptive indefinite kernels, which can better utilize the structures in the Krein space. To this end, we first propose an adaptive embedding function in the Lorentz model and define indefinite Lorentz kernels (iLks) via the embedding function. Due to the isometric relationship between the Lorentz model and the Poincare ball, these iLks are further extended to the Poincare ball, resulting in the development of what are termed indefinite Poincare kernels (iPKs). We evaluate the proposed indefinite kernels on a diversity of learning scenarios, including image classification, few-shot learning, zero-shot learning, person re-identification, knowledge distillation, etc. We show that the proposed indefinite kernels can bring significant performance gains over the baselines and enjoy better representation power from RKKSs than pd kernels.
Keyword:
Hyperbolic space
Data hierarchy
Indefinite Lorentz kernels
Indefinite Poincare kernels
期刊
IF:
6.3
论文数:
8.2K
被引数:
3.0W
机构
引用论文
Integrated Traffic and Emission Simulation: a Model Calibration Approach Using Aggregate Information
Do Perceptions of Competence Mediate The Relationship Between Fundamental Motor Skill Proficiency and Physical Activity Levels of Children in Kindergarten?能力的感知是否可以介导幼儿园儿童的基本运动技能熟练程度与身体活动水平之间的关系?
Service-Related Exposures and Physical Health Trajectories Among Aging Veteran Men老年退伍军人中与服务相关的暴露和身体健康轨迹

