When task performance deceives: Task-geometry decoupling in learnable-curvature hyperbolic GNNs
- 29 May 2026
- journal article
- Published by Elsevier in Neural Networks
- Vol. 203, 109172
Abstract
No abstract availableKeywords
Funding Information
- National Natural Science Foundation of China (71740024)
- National Social Science Fund of China (11CTQ020)
This publication has 11 references indexed in Scilit:
- Balanced Hyperbolic Embeddings Are Natural Out-of-Distribution DetectorsInternational Journal of Computer Vision, 2026
- Optimal protocols for continual learning via statistical physics and control theory*Journal of Statistical Mechanics: Theory and Experiment, 2025
- Learning Structured Representations with Hyperbolic Embeddings2024
- A control theoretic framework for adaptive gradient optimizersAutomatica, 2023
- Brain-inspired replay for continual learning with artificial neural networksNature Communications, 2020
- Microsoft Academic Graph: When experts are not enoughQuantitative Science Studies, 2020
- PID Controller-Based Stochastic Optimization Acceleration for Deep Neural NetworksIEEE Transactions on Neural Networks and Learning Systems, 2020
- Geometric Deep Learning: Going beyond Euclidean dataIEEE Signal Processing Magazine, 2017
- Riemannian Geometry1992
- The sizes of compact subsets of Hilbert space and continuity of Gaussian processesJournal of Functional Analysis, 1967