14 June - 27 June 2019

2130 new papers

Etymo Newsletter provides the latest development in machine learning research, including the most popular datasets and the most trending papers in the past two weeks.

If you like this newsletter, you can subscribe to our fortnightly newsletters here.

Fortnight Summary

Popular datasets include MNIST, ImageNet, CIFAR-10, CelebA, and COCO. Our trending phrases are Explicit Attention, Local Regression, and Online Feature. In our trending papers section, we include a paper on XLNet: Generalized Autoregressive Pretraining for Language Understanding, a paper on exchangeable stochastic processes called the Functional Neural Processes (FNPs), and a paper on an unsupervised version of capsule networks.

Popular Datasets

Here are most mentioned datasets over the last two weeks.

Name Type Number of Papers
MNIST Handwritten Digits 75
ImageNet Image Dataset 45
CIFAR-10 Tiny Image Dataset in 10 Classes 44
CelebA Large-scale CelebFaces Attributes (CelebA) Dataset 12
COCO Common Objects in Context 12

Trending Phrases

In this section, we present a list of phrases that appeared significantly more in this newsletter than the previous newsletters.

Etymo Trending

Presented below is a list of the most trending papers added in the last two weeks.

  • XLNet: Generalized Autoregressive Pretraining for Language Understanding:
    XLNet is a generalized autoregressive pretraining method that enables learning bidirectional contexts by maximizing the expected likelihood over all permutations of the factorization order and overcomes the limitations of BERT thanks to its autoregressive formulation. XLNet also integrates ideas from Transformer-XL, the state-of-the-art autoregressive model, into pretraining.

  • The Functional Neural Process:
    The Functional Neural Processes (FNPs) is a new family of exchangeable stochastic processes. FNPs model distributions over functions by learning a graph of dependencies on top of latent representations of the points in the given dataset. They are scalable to large datasets through mini-batch optimization. They can make predictions for new points via their posterior predictive distribution.

  • Stacked Capsule Autoencoders:
    The authors describe an unsupervised version of capsule networks, in which a neural encoder, which looks at all of the parts, is used to infer the presence and poses of object capsules. They learn object- and their part-capsules on unlabeled data, and then cluster the vectors of presences of object capsules.

Hope you have enjoyed this newsletter! If you have any comments or suggestions, please email or