Self-supervised surrogate tasks

[vict0rsch]

Self-Supervised Learning of Pretext-Invariant Representations PIRL

https://arxiv.org/abs/1912.01991

[vict0rsch]

Data-Efficient Image Recognition with Contrastive Predictive Coding https://arxiv.org/abs/1905.09272

arXiv.org

Data-Efficient Image Recognition with Contrastive Predictive Coding

Human observers can learn to recognize new categories of images from a handful of examples, yet doing so with machine perception remains an open challenge. We hypothesize that data-efficient recognition is enabled by representations which make the variability in natural signals more predictable. We therefore revisit and improve Contrastive Predictive Coding, an unsupervised objective for learning such representations. This new implementation produces features which support state-of-the-art linear classification accuracy on the ImageNet dataset. When used as input for non-linear classification with deep neural networks, this representation allows us to use 2-5x less labels than classifiers trained directly on image pixels. Finally, this unsupervised representation substantially improves transfer learning to object detection on PASCAL VOC-2007, surpassing fully supervised pre-trained ImageNet classifiers.

[vict0rsch]

More papers here:

https://docs.google.com/document/d/1MD9vOTvPydYuSks-xJt7Cf64ijuOcr7i2_lOmj1Jk48/edit#heading=h.w3qa5fc9vrfa

Google Docs

VICC Lit Review - May 2020

For each article, include: title abstract URL screenshot of main figure (architecture) screenshot of loss functions (if interesting) couple of sentences on why it's interesting Papers Good overview here: https://github.com/lzhbrian/image-to-image-papers/blob/master/README.md Conditional/Semantic...

[vict0rsch]

@alexrey88 if you have time, the self-supervision section will be of interest for our upcoming work

[alexrey88]

@vict0rsch cool i'll look into that :)