09.04.16 Recap — Wesley Pennington

How Tech Giants are Devising Real Ethics for Artificial Intelligence by John Markoff
Alphabet (Google), Amazon, Facebook, IBM, and Microsoft are coming together to form an industry group focused on the creation of a standard of ethics around artificial intelligence. The group's aims will be to be to ensure that AI research is focused on benefiting people, not hurting them.

One Hundred Year Study on Artificial Intelligence (AI100) by Stanford + Co. (thx Eric Horvitz)
A consortium of researchers released a report out of Stanford University that details the impact of artificial intelligence on society. The brainchild of Eric Horvitz, a computer scientist at Microsoft Research and Stanford alum, the group includes a host of leading academicians and talented contributors. Certainly an interesting read and, incidentally, the inspiration for the industry group noted earlier ☝️ .

The Next Big Shift in SaaS by Tomasz Tunguz
In addition to sharing his thoughts on the workflow of the future, which I shared a week or so ago, Tomasz discusses how SaaS applications can move from being displacers to disruptors. I had a recent conversation that arrived at a similar conclusion. Doing some reading after the fact led me to this post from July. A timely read, despite the delay.

First Round Capital's FAQs from First Round Capital
First Round Capital opens up their playbook to provide more insight into what they're looking for in new investments, their process, and how they make decisions. Great to see this type of increased transparency in the venture space. ICYMI - Bloomberg Beta recently open sourced their manual, too.

Facebook AI Research open sourcing tools
FAIR is releasing some of Facebook's tools including fastText for text classification and DeepMask, SharpMask, and MultiPathNet which are used to understand objects in photos.

A Neural Transducer from Google Brain, Google DeepMind, and OpenAI
Work a Google has developed a Neural Transducer that allows for a model to make incremental predictions as more input arrives, without redoing the entire computation.

Pydata Chicago 2016 Notes
Notes and notes from Pydata 2016 in Chicago.

Choosing the right estimator
Nice visualization from the folks at s cikit-learn - brought to my attention by Sebastian Raschka (and also featured the aforementioned Pydata notes).

Why does deep and cheap leaning work so well? by Henry W. Lin and Max Tegmark
The authors explain the mechanics of why deep learning works so well using physics.

Building Machines that Learn and Think Like People by Brenden Lake + co.
Publishing on the topic in Science last December and on arXiv earlier this year, Lake and co. developed a technique dubbed Bayesian Programming Learning (BPL) to learn from a single example, as opposed to more data hungry approaches like deep learning. The study contends that this probabilistic model is more likely how humans learn. Regardless, it is exciting as it offers a fresh take on machine intelligence which has been particularly captivated by deep learning (and for good reason).

The iBrain is Here...and it's Already Inside Your Phone by Steven Levy
A look at how Apple leverages machine learning and artificial intelligence in their products. In other news, they just acquired machine learning startup Turi and are expanding their ML/AI forces.

a16z Podcast: Pricing Free with Mark Cranney, Martin Casado, and Peter Levine
More pricing with the a16z team -- this time taking on freemium and open source models.

HN Poll: Which deep learning framework would you use?
Hacker News poll on the most popular deep learning frameworks. At the time of writing, Tensorflow is demonstrating a commanding lead with Keras (Using Theano or Tensorflow as a backend) in a distant second followed later by Theano, Caffe, and DL4J, respectfully, among others to round out the group.

Revisiting Blitzscaling w/ Reid Hoffman
ICYMI - Reid taught a class at Stanford in the Fall of 2015 on what he called Blitzscaling focused on high impact, high growth entrepreneurship. More info on the topic available here in an interview with HBR. Over the course, he had successful entrepreneurs and CEOs come in and tell their stories. Luckily, they are hosted on Youtube and I somehow found myself stumbling upon the Reed Hastings interview again recently, which is a good listen. Ran back the Eric Schmidt class, as well. Plenty of other good talks in there if you haven't heard them (or want a refresh).

Multi-scale brain networks by Rick Betzel and Danielle Bassett
Awesome review paper by my buddy, Rick, whose group also just put out another study that was profiled in MIT Tech Review: How the Mathematics of Algebraic Topology is Revolutionizing Brain Science. They're on fire 🔥

Jupiter's North Pole Unlike Anything Encountered in Solar System by NASA
Proud fan of the Jup since back in the day - even before that 2nd grade science project. Has been fun to see NASA explore.

Hope you had a good Labor Day weekend. 🎉