Deep neural decision forest (NDF) achieved remarkable performance on various vision tasks via combining decision tree and deep representation learning. In this work, we first trace the decision-making process of this model and visualize saliency maps to understand which portion of the input influence it more for both classification and regression problems. We then apply NDF on a multi-task coordinate regression problem and demonstrate the distribution of routing probabilities, which is vital for interpreting NDF yet not shown for regression problems. The pre-trained model and code for visualization will be available at https://github.com/Nicholasli1995/VisualizingNDF

Segmentation of pancreas is important for medical image analysis, yet it faces great challenges of class imbalance, background distractions and non-rigid geometrical features. To address these difficulties, we introduce a Deep Q Network(DQN) driven approach with deformable U-Net to accurately segment the pancreas by explicitly interacting with contextual information and extract anisotropic features from pancreas. The DQN based model learns a context-adaptive localization policy to produce a visually tightened and precise localization bounding box of the pancreas. Furthermore, deformable U-Net captures geometry-aware information of pancreas by learning geometrically deformable filters for feature extraction. Experiments on NIH dataset validate the effectiveness of the proposed framework in pancreas segmentation.

A Japanese tech start-up is using deep learning to teach a pair of machines a simple job for a human, but a surprisingly tricky task for a robot - cleaning a bedroom. Though it may seem like a basic, albeit tedious, task for a human, robots find this type of job surprisingly complicated. A Japanese tech start-up is using deep learning to teach AI how to deal with disorder and chaos in a child's room. Deep learning is where algorithms, inspired by the human brain, learn from large amounts of data so they're able to perform complex tasks. Some tasks, like welding car chassis in the exact same way day after day, are easy for robots as it is a repetitive process and the machines do not suffer with boredom in the same way as disgruntled employees.

The New York Times has confirmed what some have long suspected: The Chinese government is using a "vast, secret system" of artificial intelligence and facial recognition technology to identify and track Uighurs--a Muslim minority, 1 million of whom are being held in detention camps in China's northwest Xinjiang province. This technology allows the government to extend its control of the Uighur population across the country. It may seem difficult to imagine a similar scenario in the U.S., but related technologies, built by Amazon, are already being used by U.S. law enforcement agencies to identify suspects in photos and video. And echoes of China's system can be heard in plans to deploy these technologies at the U.S.-Mexico border. A.I. systems also decide what information is presented to you on social media, which ads you see, and what prices you're offered for goods and services.

The UK is about to introduce restrictions on watching pornography of a kind never before seen in the world. The government is planning to stop children being damaged by watching adult content by stopping anyone from doing so unless they go through a "rigorous" age verification process. Websites that aren't part of the blocks could find themselves being punished or blocked entirely within the UK. We'll tell you what's true. You can form your own view.

The recently announced identity checks to stop under-18s from visiting pornographic websites in the UK have led to a surge in interest in technology that would allow people to bypass them. Critics claim the new rules are "disastrous" for people's privacy and are fundamentally flawed due to the ease of which they can be circumvented using virtual private networks (VPNs). Searches for VPNs on Google's search engine tripled in the hours following the government's announcement that the verification system would come into effect in July. We'll tell you what's true. You can form your own view.

Smart lights still tend to offer a rude awakening if you tie them to your alarm, but Google wants to fix that. It's delivering a promised Gentle Sleep & Wake feature for Home speakers that gradually dims or brightens your Philips Hue lights to provide a more natural rest. Say the right command (such as "turn on Gentle Wake Up," "wake up my lights" or "sleep my lights") and the lights will change over the course of half an hour. You can set specific times if you intend to use it as part of your daily routine. The feature won't be widely available, at least not for a while.

Isn't it true that we are living in a digitalized world that has eliminated tons of human work by positioning automation?. In fact, it is the most defined period as Google's self-driving car has been invented. But, this period is not in its final stages instead is multiplying to create many more awesome things to surface in the near future. The most exciting concept that sits beside all these major transformations is Machine Learning, which is nothing but allowing computers to learn on their own to arrive at useful insights. Supervised learning is similar to a teacher teaching his students with examples and after sufficient practice, the teacher stops supervising and let the students derive at their own solution.

TEL AVIV, Israel, 16 April 2019--Israel's reams of electronic medical records –health data on its population of around 8.9 million people-- are proving fruitful for a growing number of digital health startups training algorithms to do things like early detection of diseases and produce more accurate medical diagnoses. According to a new report by Start-Up Nation Central, the growth in the number of Israeli digital health startups –537 companies, up from 327 in 2014--has drawn in new investors, including Israeli VCs who have never previously invested in healthcare. This has driven financing in the sector to a record $511M in 2018, up 32% year on year. By the first quarter of 2019 the amount raised was already at $214M. Of the $511M, over 50% ($285M) went to companies in decision support and diagnostics which rely heavily on data crunching.

DScribe is a software package for machine learning that provides popular feature transformations ("descriptors") for atomistic materials simulations. DScribe accelerates the application of machine learning for atomistic property prediction by providing user-friendly, off-the-shelf descriptor implementations. The package currently contains implementations for Coulomb matrix, Ewald sum matrix, sine matrix, Many-body Tensor Representation (MBTR), Atom-centered Symmetry Function (ACSF) and Smooth Overlap of Atomic Positions (SOAP). Usage of the package is illustrated for two different applications: formation energy prediction for solids and ionic charge prediction for atoms in organic molecules. The package is freely available under the open-source Apache License 2.0.