Machine Learning Weekly Issue #01
Facebook using AI to improve photo descriptions, using machine learning to predict hard-drive failures, and many more.
Welcome to the machine learning weekly section of my publications. This section is a weekly, sometimes bi-weekly blog with the most important and interesting stories in the machine learning field. You can follow the blog for free.
Schedule of new articles
I manage the entire blog all by myself. I intend to post Monday morning every week, but if I'm busy with university or something then I'll delay this delivery by a week.
Welcome to the first issue of machine learning weekly. I worked tirelessly throughout the week to find you the best of the best research in machine learning and the field of AI.
How Facebook is using AI to improve photo descriptions for people who are blind or visually impaired.
Official Article here
When Facebook users scroll through their News Feed, they find all kinds of content — articles, friends’ comments, event invitations, and of course, photos. Most people are able to instantly see what’s in these images, whether it’s their new grandchild, a boat on a river, or a grainy picture of a band on stage. But many users who are blind or visually impaired (BVI) can also experience that imagery, provided it’s tagged properly with alternative text (or “alt text”). A screen reader can describe the contents of these images using a synthetic voice and enable people who are BVI to understand images in their Facebook feed. If you are into machine learning, it's possible that you'll like my articles with in-depth articles and weekly deep learning updates too.
Every now and then your hard drive fails. Predicting hard drive failure with machine learning
Official article here
We’ve all had a hard drive fail on us, and often it’s as sudden as booting your machine and realizing you can’t access a bunch of your files. It’s not a fun experience. It’s especially not fun when you have an entire data center full of drives that are all important to keeping your business running. What if we could predict when one of those drives would fail, and get ahead of it by preemptively replacing the hardware before the data is lost? This is where the history of predictive drive failure at Datto begins.
In case you missed: The Unsplash Dataset, the world’s largest open library dataset for free.
Find the unsplash dataset here. Train and test models using the largest collaborative image dataset ever openly shared. The Unsplash Dataset is created by over 200,000 contributing photographers and billions of searches across thousands of applications, uses, and contexts. They have 2 different dataset versions, 1st is the lite version and another one is the full dataset. The Lite version has over 25000+ images (550 MB) and the full version has 2,000,000+ images (25GB, non-commercial usage only).
DeepDT: Learning Geometry From Delaunay Triangulation for Surface Reconstruction
Find the pdf here
In this paper, a novel learning-based network, named DeepDT, is proposed to reconstruct the surface from Delaunay triangulation of the point cloud. DeepDT learns to predict inside/outside labels of Delaunay tetrahedrons directly from a point cloud and corresponding Delaunay triangulation. The local geometry features are first extracted from the input point cloud and aggregated into a graph deriving from the Delaunay triangulation. Then a graph filtering is applied to the aggregated features in order to add structural regularization to the label prediction of tetrahedrons. Due to the complicated spatial relations between tetrahedrons and the triangles, it is impossible to directly generate ground truth labels of tetrahedrons from the ground truth surface. Here researchers proposed a multilabel supervision strategy that votes for the label of a tetrahedron with labels of sampling locations inside it. The proposed DeepDT can maintain abundant geometry details without generating overly complex surfaces , especially for inner surfaces of open scenes. Meanwhile, the generalization ability and time consumption of the proposed method is acceptable and competitive compared with the state-of-the-art methods. Experiments demonstrate the superior performance of the proposed DeepDT.