Four short links: 12 June 2017

By June 12, 2017ai, bigdata, machinelearning

Modern Web Spellbook, GPU Gap, Measure What Matters, and Educational Robotics Toy

  1. Spellbook of Modern Web DevThis document originated from a bunch of most-commonly used links and learning resources I sent to every new web developer on our full-stack web development team. For each problem domain and each technology, I try my best to pick only one or a few links that are most important, typical, common, or popular and not outdated, based on clear trends, public data, and empirical observation.
  2. How AI Can Keep Accelerating After Moore’s Law — answer: GPUs and innovation therein. Nvidia CEO Jensen Huang displayed a chart showing how his chips’ performance has continued to accelerate exponentially while growth in the performance of general purpose processors, or CPUs, has slowed. Doug Burger, a distinguished engineer at Microsoft’s NExT division that works on commercializing new technology, says a similar gap is opening between conventional and machine learning software. “You’re starting to see a [performance] plateau for general software—it has stopped improving at historical rates—but this AI stuff is still increasing rapidly,” he says. Also: Google’s machine learning how-to-optimize-machine-learning result would cost you $250K to reproduce on Amazon GPUs.
  3. Gamified Wikipedia Tutorial Didn’t Change Participation Rates (Benjamin Mako Hill) — To our surprise, we found that, in both cases, there were no significant effects on any of the outcomes of interest. Being invited to play the Wikipedia Adventure, therefore, had no effect on new users’ volume of participation either on Wikipedia in general, or on talk pages specifically, nor did it have any effect on the average quality of edits made by the users in our study. Despite the very positive feedback that the system received in the survey evaluation stage, it did not produce a significant change in newcomer contribution behavior. We concluded that the system by itself could not reverse the trend of newcomer attrition on Wikipedia. A reminder that you should, as Mako Hill did, measure the behaviour you care about, not how much people enjoyed your intervention.
  4. Sony Toiothe result of five years of research into developing a toy that’s simple enough for kids to use, but also sophisticated enough to create a figurative sandbox where kids can explore the inner workings of robotics engineering.

Continue reading Four short links: 12 June 2017.

Source link