5th April - 22nd April 2019

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

Here are most mentioned datasets over the last two weeks.

Trending Phrases

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

Generative Map:
• A Generative Map for Image-based Camera Localization

Deep Local:
• Revisiting Local Descriptor based Image-to-Class Measure for Few-shot Learning
• Needle in a Haystack: A Framework for Seeking Small Objects in Big Datasets

Relation Graph:
• A Structured Model For Action Detection
• Few-shot Learning: A Survey

Etymo Trending

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

• Approximation and Non-parametric Estimation of ResNet-type Convolutional Neural Networks:
  "Convolutional neural networks (CNNs) are unrealistically wide and difficult to obtain via optimization due to sparse constraints in important function classes, including the H¨older class." This paper shows "a ResNet-type CNN can attain the minimax optimal error rates in these classes."


• Precise Detection in Densely Packed Scenes:
  This paper proposes a novel, deep-learning based method for precise object detection, i.e., man-made scenes that contains numerous objects, often identical, positioned in close proximity.


• Fairwashing: the risk of rationalization:
  Black-box explanation includes model explanation, outcome explanation, and model inspection. This paper shows that these techniques can be used in a negative manner to perform fairwashing. This paper proposes a method for searching for fair rule lists approximating an unfair black-box model.

Easter Holiday Special: Latest in Knowledge Graph

• One for All: Neural Joint Modeling of Entities and Events:
  This paper introduces a model that can jointly perform predictions for entity mentions, event triggers and arguments using shared hidden representations.


• Entity Alignment between Knowledge Graphs Using Attribute Embeddings:
  This paper proposes a model that exploits large numbers of attribute triples in knowledge graph and generates attibute character embeddings in order to help align entities in different knowledge graphs.


• SimplE Embedding for Link Prediction in Knowledge Graphs:
  This paper presents a simple enhancement of Canonical Polyadic (CP) decomposition, a well-known method for link predicitons on knowledge graph. The code is available at https://github.com/Mehran-k/SimplE.


• KADE : Aligning Knowledge Base and Document Embedding Models using Regularized MultiTask Learning?:
  Text documents contain freeform entities while knowledge bases (KB) organize entities in a structured way. This paper proposes a method to represent embeddings by different models in a common vector space, while preserving the semantics of the embedding models it integrates.


• Formal Query Generation for Question Answering over Knowledge Bases:
  This paper proposes a SPARQL query generator, enabling easy integration with other components for the construction of a fully functional question answering (QA) system.

Adapted from PaperWeekly

Hope you have enjoyed this newsletter! If you have any comments or suggestions, please email ernest@etymo.io or steven@etymo.io.