|July 2 · Issue #46 · View online |
Howdy and welcome to another exciting week in Deep Learning.
| Nvidia partners with Volvo to power self-driving cars - Business Insider |
Another round in the race to the self-driving car; NVIDIA announced Monday it will partner with Volvo to power self-driving cars by 2021.
| DeepMind Health inks another 5-year NHS app deal in face of ongoing controversy |
DeepMind Health, the division of the Google-owned AI company that’s applying machine learning to medical data in the hopes of profiting from diagnostic gain, has inked another services agreement with the U.K.’s National Health Service — expanding the deployment of an alerts, messaging and task management app, Streams, to a hospital in Taunton & Somerset.
| AI Progress Measurement | Electronic Frontier Foundation |
Really cool project which collects problems and metrics/datasets from the AI research literature, and tracks progress on them: ‘You can use this notebook to see how things are progressing in specific subfields or AI/ML as a whole, as a place to report new results you’ve obtained, as a place to look for problems that might benefit from having new datasets/metrics designed for them, or as a source to build on for data science projects.’
| Learning to Reason with Neural Module Networks |
Fascinating blog post introducing neural module networks (NMNs) which incorporate more flexible approaches to problem-solving while preserving the expressive power that makes deep learning so effective.
| Predicting the Development of Lung Cancer from CT Images |
Great write about by Julian de Wit about how he and Daniel Hammack went about dissecting the problem of detecting lung cancer from lung CT scans to finally arrive at a clever solution employing CNNs.
| An Overview of Multi-Task Learning for Deep Learning |
| Dexterity Network 2.0 - Dataset for Deep Grasping |
The Dexterity Network (Dex-Net) 2.0 is a project centered on using physics-based models of robust robot grasping to generate massive datasets of parallel-jaw grasps across thousands of 3D CAD object models. These datasets are used to train deep neural networks to plan grasps from a point clouds on a physical robot that can lift and transport a wide variety of objects.
| DeepMind Research – Kinetics Dataset |
Kinetics is a large-scale, high-quality dataset of YouTube video URLs which include a diverse range of human focused actions. The dataset consists of approximately 300,000 video clips, and covers 400 human action classes with at least 400 video clips for each action class. Each clip lasts around 10s and is labeled with a single class.
| GitHub - webdnn: Fastest DNN Execution Framework on Web Browser |
WebDNN is an open source software framework for executing deep neural network (DNN) pre-trained model on web browser.
| PixelGAN Autoencoders |
“PixelGAN autoencoder” is a generative autoencoder in which the generative path is a convolutional autoregressive neural network on pixels (PixelCNN) that is conditioned on a latent code, and the recognition path uses a generative adversarial network (GAN) to impose a prior distribution on the latent code.
| The Marginal Value of Adaptive Gradient Methods in Machine Learning |
This paper shows that adaptive optimization methods, which perform local optimization with a metric constructed from the history of iterates, such as AdaGrad, RMSProp, and Adam often find drastically different solutions than gradient descent (GD) or stochastic gradient descent (SGD).