Geoopt: Riemannian Optimization in PyTorch
arXiv:2005.02819
Abstract
Geoopt is a research-oriented modular open-source package for Riemannian Optimization in PyTorch. The core of Geoopt is a standard Manifold interface that allows for the generic implementation of optimization algorithms. Geoopt supports basic Riemannian SGD as well as adaptive optimization algorithms. Geoopt also provides several algorithms and arithmetic methods for supported manifolds, which allow composing geometry-aware neural network layers that can be integrated with existing models.
Proceedings of the 37th International Conference on Machine Learning, Vienna, Austria, PMLR 108, 2020, GRLB Workshop
References in corpus (13)
- PyTorch: An Imperative Style, High-Performance Deep Learning Library
- Fast Graph Representation Learning with PyTorch Geometric
- Manopt, a Matlab toolbox for optimization on manifolds
- Unitary Evolution Recurrent Neural Networks
- Poincaré Embeddings for Learning Hierarchical Representations
- Hyperbolic Graph Neural Networks
- Riemannian batch normalization for SPD neural networks
- Hyperbolic Neural Networks
- Manifold Optimization Over the Set of Doubly Stochastic Matrices: A Second-Order Geometry
- Probabilistic Permutation Synchronization using the Riemannian Structure of the Birkhoff Polytope
- McTorch, a manifold optimization library for deep learning
- Orthogonal Recurrent Neural Networks with Scaled Cayley Transform
- Hyperbolic Interaction Model For Hierarchical Multi-Label Classification
Cited by in corpus (15)
- Graph Neural Networks on SPD Manifolds for Motor Imagery Classification: A Perspective from the Time-Frequency Analysis
- From Trees to Continuous Embeddings and Back: Hyperbolic Hierarchical Clustering
- Manifolds.jl: An Extensible Julia Framework for Data Analysis on Manifolds
- Hyperbolic Deep Neural Networks: A Survey
- A contribution to Optimal Transport on incomparable spaces
- Symmetric Spaces for Graph Embeddings: A Finsler-Riemannian Approach
- Manifolds for Unsupervised Visual Anomaly Detection
- Fully Hyperbolic Rotation for Knowledge Graph Embedding
- Hyperbolic Neural Networks++
- HypLL: The Hyperbolic Learning Library
- Disentangling images with Lie group transformations and sparse coding
- Riemannian Functional Map Synchronization for Probabilistic Partial Correspondence in Shape Networks
- Identifying and Mitigating Gender Bias in Hyperbolic Word Embeddings
- HyperExpan: Taxonomy Expansion with Hyperbolic Representation Learning
- Extracting Event Temporal Relations via Hyperbolic Geometry