All
All

What are you looking for?

All
Projects
Organizations

Quick search

  • Projects supported by TA ČR
  • Excellent projects
  • Projects with the highest public support
  • Current projects

Smart search

  • That is how I find a specific +word
  • That is how I leave the -word out of the results
  • “That is how I can find the whole phrase”

H-NeXt: The next step towards roto-translation invariant networks

Result description

The widespread popularity of equivariant networks underscores the significance of parameter efficient models and effective use of training data. At a time when robustness to unseen deformations is becoming increasingly important, we present H-NeXt, which bridges the gap between equivariance and invariance. H-NeXt is a parameter-efficient roto-translation invariant network that is trained without a single augmented image in the training set. Our network comprises three components: an equivariant backbone for learning roto-translation independent features, an invariant pooling layer for discarding roto-translation information, and a classification layer. H-NeXt outperforms the state of the art in classification on unaugmented training sets and augmented test sets of MNIST and CIFAR-10

Keywords

H-NeXTrobustness to unseen deformationsparameter-efficient roto-translation invariant networkclassification on unaugmented training set

The result's identifiers

Alternative languages

  • Result language

    angličtina

  • Original language name

    H-NeXt: The next step towards roto-translation invariant networks

  • Original language description

    The widespread popularity of equivariant networks underscores the significance of parameter efficient models and effective use of training data. At a time when robustness to unseen deformations is becoming increasingly important, we present H-NeXt, which bridges the gap between equivariance and invariance. H-NeXt is a parameter-efficient roto-translation invariant network that is trained without a single augmented image in the training set. Our network comprises three components: an equivariant backbone for learning roto-translation independent features, an invariant pooling layer for discarding roto-translation information, and a classification layer. H-NeXt outperforms the state of the art in classification on unaugmented training sets and augmented test sets of MNIST and CIFAR-10

  • Czech name

  • Czech description

Classification

  • Type

    O - Miscellaneous

  • CEP classification

  • OECD FORD branch

    20206 - Computer hardware and architecture

Result continuities

Others

  • Publication year

    2023

  • Confidentiality

    S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Basic information

Result type

O - Miscellaneous

O

OECD FORD

Computer hardware and architecture

Year of implementation

2023