|August 30 · Issue #4 · View online |
Issue 4 of DL weekly you’ll find how Apple uses deep learning internally, how facebook AI research proposes to segment and classify objects in images, and how to use neural networks to compress images,
happy reading and hopefully hacking!
| An Exclusive Look at How AI and Machine Learning Work at Apple |
“Yes, there is an ‘Apple brain’ - it’s already inside your iPhone.” Apple uses deep learning to detect fraud and identify most useful feedback from its beta testers.
| Nigel Artificial Intelligence Learns Common Sense Via Observation |
In an AI industry first, Kimera System’s Nigel artificial general intelligence technology applies common sense actions to relevant situations it determined through unsupervised learning and observation.
| Reinforcement Learning and DQN, learning to play from pixels |
Interesting introduction to reinforcement learning accompanied by a detailed explanation of DQN (Deep Q-Network) for pixel inputs.
| RNNs in Tensorflow, a Practical Guide and Undocumented Features – WildML |
Extremely valuable tutorial by Danny Britz for learning some of the best practices for working with RNNs in Tensorflow, especially since he explores functionality that isn’t well documented on the official site.
| Deep Learning: The Mental Model |
Good beginner’s intro on how to think about the way ML problems are tackled with Deep Learning.
| Machine Learning is Fun Part 5: Language Translation with Deep Learning and the Magic of Sequences |
Part 5 of Adam Geitgey’s ML series, explores Deep Learning using sequence-to-sequence learning in machine translation as an example.
| Deep Learning: Definition, Resources, Comparison with Machine Learning - Data Science Central |
Some much-needed clarity on how to properly delineate deep learning from other fields of machine learning an AI.
| Deep Learning Part 2: Transfer Learning and Fine-tuning Deep Convolutional Neural Networks |
Fascinating and thorough post by Anusua Trivedi, Microsoft Data Scientist, on Deep Convolutional Neural Networks (DCNNs) and how Transfer learning and Fine-tuning helps better the training process for domain specific images.
| AnthemScore - Music Transcription Software |
“It’s all signals, Jerry, it’s signals!” Perhaps unsurprisingly this post finds that detection in music is essentially image recognition with a few small differences.
| Segmenting and refining images with SharpMask |
Facebook AI Research dropped another one this week; they introduce an interesting three step process where one network, DeepMask, segments the images another, SharpMask, refines them to identify sharp object boundaries and a third, Region-CNN, classifies the segmented objects.
| GitHub - david-gpu/srez: Image super-resolution through deep learning |
srez - Image super-resolution through deep learning
| Full Resolution Image Compression with Recurrent Neural Networks |
This paper presents a set of full-resolution lossy image compression methods based on neural networks.
| Stein Variational Gradient Descent: A General Purpose Bayesian Inference Algorithm |
This paper proposes a general purpose variational inference algorithm that forms a natural counterpart of gradient descent for optimization.