|April 17 · Issue #105 · View online |
If there’s anything you think we missed or want to see in next week’s issue, send us a DM on Twitter: @dl_weekly
Until next week!
| This AI-generated sculpture is made from the shredded remains of the computer that designed it - The Verge |
Art made with the help of artificial intelligence is a growing scene. New York artist Ben Snell has created Dio — an artwork that might be the first piece of AI-generated sculpture to go up for auction. Dio was created by algorithms and cast using the ground up remains of the computer used to design it.
| Google launches an end-to-end AI platform – TechCrunch |
Google launched an end-to-end AI platform that allows users to perform every step in their AI workflow. From data ingestion and labeling to model training and deployment. AutoML also expands to tabular data and video.
| Disney's AI generates storyboard animations from scripts | VentureBeat |
Researchers at Disney and Rutgers propose in a newly published paper an AI system that can generate animations from screenplay snippets.
| A new bill would force companies to check their algorithms for bias - The Verge |
Senators Cory Booker (D-NJ) and Ron Wyden (D-OR), as well as Rep. Yvette Clarke (D-NY), have introduced the “Algorithmic Accountability Act” — which would require companies to audit their AI-based tools for discrimination.
| One Month, 500,000 Face Scans: How China Is Using A.I. to Profile a Minority - The New York Times |
In a major ethical leap for the tech world, Chinese start-ups have built algorithms that the government uses to track members of a largely Muslim minority group.
| Smart home, machine learning and discovery — Benedict Evans |
Insightful thoughts on how machine learning may end up woven into our lives and homes.
| Deep Learning in Production: Sentiment Analysis with the Transformer Model |
In 2017, the Google Brain team published Attention Is All You Need, a paper that studied the effectiveness of attention mechanisms in neural networks. At a time when recurrent and convolutional…
| How to Get Started With Deep Learning for Computer Vision (7-Day Mini-Course) |
A free mini-course to get you started with deep learning for computer vision using Keras and the rest of the scientific Python stack.
| GitHub - fritzlabs/Awesome-Mobile-Machine-Learning: A curated list of awesome mobile machine learning resources. |
A curated list of awesome mobile machine learning resources. - fritzlabs/Awesome-Mobile-Machine-Learning
| GitHub - PolyAI-LDN/conversational-datasets: Datasets for conversational AI |
Datasets for conversational AI. Contribute to PolyAI-LDN/conversational-datasets development by creating an account on GitHub.
| MLIR: A new intermediate representation and compiler framework |
The TensorFlow team unveils a new intermediate representation for machine learning that will make it much easier for ML pipelines to target heterogenous hardware.
| GitHub - Omegastick/pytorch-cpp-rl: PyTorch C++ Reinforcement Learning |
| GitHub - pytorch/ignite: High-level library to help with training neural networks in PyTorch |
High-level library to help with training neural networks in PyTorch .
| Automatic adaptation of object detectors to new domains using self-training |
Abstract: This work addresses the unsupervised adaptation of an existing object detector to a new target domain. We assume that a large number of unlabeled videos from this domain are readily available. We automatically obtain labels on the target data by using high-confidence detections from the existing detector, augmented with hard (misclassified) examples acquired by exploiting temporal cues using a tracker. These automatically-obtained labels are then used for re-training the original model….
| [1904.05049] Drop an Octave: Reducing Spatial Redundancy in Convolutional Neural Networks with Octave Convolution |
Abstract: ….In this work, we propose to factorize the mixed feature maps by their frequencies and design a novel Octave Convolution (OctConv) operation to store and process feature maps that vary spatially “slower” at a lower spatial resolution reducing both memory and computation cost. Unlike existing multi-scale methods, OctConv is formulated as a single, generic, plug-and-play convolutional unit that can be used as a direct replacement of (vanilla) convolutions without any adjustments in the network architecture….We experimentally show that by simply replacing convolutions with OctConv, we can consistently boost accuracy for both image and video recognition tasks, while reducing memory and computational cost. An OctConv-equipped ResNet-152 can achieve 82.9% top-1 classification accuracy on ImageNet with merely 22.2 GFLOPs.
| A Song of Ice and Data - Predicting GoT character deaths |
Studies use neural networks and bayesian survival analysis to predict the likelihood of death for Game of Thrones characters. WARNING: The main site has no spoilers, but if you click around long enough, you might find some.
For more deep learning news, tutorials, code, and discussion, join us on Slack, Twitter, and GitHub.