Digests » 103
this week's favorite
Deep neural network (DNN) is an indispensable machine learning tool for achieving human-level performance on many learning tasks. Yet, due to its black-box nature, it is inherently difficult to understand which aspects of the input data drive the decisions of the network. There are various real-world scenarios in which humans need to make actionable decisions based on the output DNNs. Such decision support systems can be found in critical domains, such as legislation, law enforcement, etc.
Geometric deep learning (GDL) is one of many advances in computer vision sparked by performance improvements in DL. This article covers an introduction to Geometric Deep Learning, its interpretation in the context of "relational inductive bias" (a term coined by DeepMind's researchers in the field), and several of many interesting use-cases (including graph segmentation, graph classification, and KGCNs). The article is also loaded with relevant references for those interested in delving deeper.
We present an algorithm for reconstructing dense, geometrically consistent depth for all pixels in a monocular video. We leverage a conventional structure-from-motion reconstruction to establish geometric constraints on pixels in the video.
Three months ago, I participated in a data science challenge that took place at my company. The goal was to help a marine researcher better identify whales based on the appearance of their flukes. More specifically, we were asked to predict for each image of a test set, the top 20 most similar images from the full database (train+test). This was not a standard classification task.
A cheat sheet for busy ML practitioners who need to run numerous modeling experiments quickly in a tidy Jupyter workspace.