Web Picks (week of 11 June 2018)

Every two weeks, we find the most interesting data science links from around the web and collect them in Data Science Briefings, the DataMiningApps newsletter. Subscribe now for free if you want to be the first to get up to speed on interesting resources.

  • AI at Google: our principles
    “We acknowledge that this area is dynamic and evolving, and we will approach our work with humility, a commitment to internal and external engagement, and a willingness to adapt our approach as we learn over time.”
  • Does China’s digital police state have echoes in the West?
    The state can gather more information, more easily, than ever before. Do not underestimate the risks
  • Facebook confirms data sharing with Chinese companies
    “Facebook Inc (FB.O) said Tuesday it has data sharing partnerships with at least four Chinese companies including Huawei, the world’s third largest smartphone maker, which has come under scrutiny from U.S. intelligence agencies on security concerns.”
  • Facebook Gave Device Makers Deep Access to Data on Users and Friends
    The company formed data-sharing partnerships with Apple, Samsung and dozens of other device makers, raising new concerns about its privacy protections.
  • Facebook let select companies have “special access” to user data
    Such data sharing was supposed to have been fully cut off in 2015, but it wasn’t.
  • Import AI has an interesting timeline of dubious things that people have synthesized via AI
    “Think fake news is bad now? ‘Deep Video Portraits’ will make it much, much worse…”
  • Attacks against machine learning — an overview
    This blog post survey the attacks techniques that target AI (artificial intelligence) systems and how to protect against them.
  • More data and surveillance are transforming justice systems
    The relationship between information and crime has changed.
  • Footsteps, Pressure Sensors, and AI: The Next Step in Airport Security
    “With this system, the researchers claim that the way a person walks and analysis of that individual’s footsteps could be used as a biometric at airport security instead of fingerprinting and eye-scanning, providing a non-intrusive method of identity verification.”
  • Chatbots were the next big thing: what happened?
    “Our hopes were sky high. Bright-eyed and bushy-tailed, the industry was ripe for a new era of innovation: it was time to start socializing with machines.”
  • Polyaxon, a platform for building, training and monitoring large scale deep learning applications.
    Looks very interesting: an open source platform for reproducible machine learning at scale.
  • RL with PyTorch
    Notebooks describing actor critic, proximal policy optimization, acer, ddpg, twin dueling ddpg, soft actor critic, generative adversarial imitation learning, and hindsight experience replay.
  • WTTE-RNN a framework for churn and time to event prediction
    “A less hacky machine-learning framework for churn- and time to event prediction. Forecasting problems as diverse as server monitoring to earthquake- and churn-prediction can be posed as the problem of predicting the time to an event. WTTE-RNN is an algorithm and a philosophy about how this should be done.”
  • Playing Atari with Six Neurons (paper)
    “We also introduce new techniques allowing both the neural network and the evolution strategy to cope with varying dimensions. This enables networks of only 6 to 18 neurons to learn to play a selection of Atari games with performance comparable—and occasionally superior—to state-of-the-art techniques using evolution strategies on deep networks two orders of magnitude larger.”
  • Relational inductive biases, deep learning, and graph networks (paper)
    “We argue that combinatorial generalization must be a top priority for AI to achieve human-like abilities, and that structured representations and computations are key to realizing this objective. Just as biology uses nature and nurture cooperatively, we reject the false choice between “hand-engineering” and “end-to-end” learning, and instead advocate for an approach which benefits from their complementary strengths. We explore how using relational inductive biases within deep learning architectures can facilitate learning about entities, relations, and rules for composing them. We present a new building block for the AI toolkit with a strong relational inductive bias–the graph network–which generalizes and extends various approaches for neural networks that operate on graphs, and provides a straightforward interface for manipulating structured knowledge and producing structured behaviors. We discuss how graph networks can support relational reasoning and combinatorial generalization, laying the foundation for more sophisticated, interpretable, and flexible patterns of reasoning.”
  • A Course in Machine Learning
    A free ebook in progress.