DeepAI AI Chat
Log In Sign Up

Equivariant Representations for Non-Free Group Actions

by   Luis Armando Pérez Rey, et al.
KTH Royal Institute of Technology
TU Eindhoven

We introduce a method for learning representations that are equivariant with respect to general group actions over data. Differently from existing equivariant representation learners, our method is suitable for actions that are not free i.e., that stabilize data via nontrivial symmetries. Our method is grounded in the orbit-stabilizer theorem from group theory, which guarantees that an ideal learner infers an isomorphic representation. Finally, we provide an empirical investigation on image datasets with rotational symmetries and show that taking stabilizers into account improves the quality of the representations.


Equivariant Representation Learning via Class-Pose Decomposition

We introduce a general method for learning representations that are equi...

Homomorphism Autoencoder – Learning Group Structured Representations from Observed Transitions

How can we acquire world models that veridically represent the outside w...

Cryptographic Group and Semigroup Actions

We consider actions of a group or a semigroup on a set, which generalize...

Pre-Learning Environment Representations for Data-Efficient Neural Instruction Following

We consider the problem of learning to map from natural language instruc...

Bispectral Neural Networks

We present a novel machine learning architecture, Bispectral Neural Netw...

Free actions and Grassmanian variety

An algebraic notion of representational consistency is defined. A theore...

Left ideal LRPC codes and a ROLLO-type cryptosystem based on group algebras

In this paper we introduce left ideal low-rank parity-check codes by usi...