Morpho-MNIST: quantitative assessment and diagnostics for representation learning

MaRDI QIDQ5214288zbMATH OpenFDO

Authors Daniel Castro, Jeremy Tan, Bernhard Kainz, Ender Konukoglu, Ben Glocker

Publication date 7 February 2020

Full work available at URL https://arxiv.org/abs/1809.10780, http://jmlr.csail.mit.edu/papers/v20/19-033.html

zbMATH Keywords

disentanglementempirical evaluationgenerative modelsmorphometricsrepresentation learning

Mathematics Subject Classification ID

Classification and discrimination; cluster analysis (statistical aspects) (62H30) Image analysis in multivariate analysis (62H35) Software, source code, etc. for problems pertaining to computer science (68-04) Learning and adaptive systems in artificial intelligence (68T05)

Recommendations

• Manifold adversarial training for supervised and semi-supervised learning
• Approximating morphological operators with part-based representations learned by asymmetric auto-encoders
• Unsupervised discovery, control, and disentanglement of semantic attributes with applications to anomaly detection
• Deformable classifiers
• Unsupervised learning of invariant representations

Cites work

• A kernel two-sample test
• An Introduction to Variational Autoencoders
• Comments on the ``Core vector machines: fast SVM training on very large data sets
• scientific article; zbMATH DE number 837911 (Why is no real title available?)
• Human-level concept learning through probabilistic program induction
• Robust vision-based features and classification schemes for off-line handwritten digit recognition

Cited in

(2)

• Morpho-MNIST
• Unsupervised discovery, control, and disentanglement of semantic attributes with applications to anomaly detection

Describes a project that uses

Uses Software

• MNIST
• InfoGAN
• Omniglot
• scikit-image
• affNIST
• EMNIST
• dSprites
• Fashion-MNIST
• GitHub

This page was built for publication: Morpho-MNIST: quantitative assessment and diagnostics for representation learning

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5214288)