I grew up being a Bayesian, apparently

Little did I know when I started my career as a research physicist at CERN that I was a member of the Bayesianist “tribe”. In fact, I was not even aware back then that what we called data analysis “another day working with data” was even a branch of the Machine Learning religion.

The content below is from the The Master Algorithm by Pedro Domingos. Formatting all mine.

Tribe Premise Master Algorithm
Symbolists All intelligence can be reduced to manipulating symbols Inverse Deduction: It figures out what knowledge is missing in order to make a deduction go through, and then makes it as general as possible
Connectionists Learning is what the brain does, and we need to reverse engineer it Back Propagation: It compares a system’s output with the desired one and then successively changes the connections in layer after layer of neurons so as to bring the output closer to what it should be
Evolutionaries The mother of all learning is natural selection Genetic programming: It mates and evolves computer programs in the same way that nature mates and evolves organisms
Bayesians All learned knowledge is uncertain, and learning itself is a form of uncertain inference Bayes’ theorem: It tells us how to incorporate new evidence into our beliefs, and probabilistic inference algorithms do that as efficiently as possible
Analogizers The key to learning is recognizing similarities between situations and thereby inferring other similarities. Support vector machine: It figures out which experiences to remember and how to combine them to make new predictions

AI bots that create their own language

I found this fascinating: Cade Metz for Wired:

As detailed in a research paper published by OpenAI this week, Mordatch and his collaborators created a world where bots are charged with completing certain tasks, like moving themselves to a particular landmark. The world is simple, just a big white square—all of two dimensions—and the bots are colored shapes: a green, red, or blue circle. But the point of this universe is more complex. The world allows the bots to create their own language as a way collaborating, helping each other complete those tasks.

You could thing about this as a new level of Cryptophasia, i.e. language created by twins that only the two children can understand. Some might say that it is scary, some might say that it is amazing that we are getting to this level of reinforced learning.

Didi has opened a self-driving lab in Mountain View

Johana Bhuiyan for Recode:

The Chinese company’s new U.S. lab, which will focus on intelligent driving systems and AI-based security for transportation, also formalizes what many already knew: Didi is working on self-driving cars.

The company has already partnered with Udacity — a college-level nanodegree startup — on its self-driving program, at the end of which Didi and a number of other partnering companies get first pick of the graduates the companies want to hire.

Didi did not only acquired Uber China assets last year but it is also actively poaching AV talent from Google Waymo and Uber itself.

I can just but imagine the potential of AVs deployed at a Didi scale in China in a near future.

Alphabet vs Uber

Julia Love and Heather Somerville for Reuters:

Now, if the Waymo suit damages Uber, GV’s investment in the ride-hailing company stands to go down as a Silicon Valley rarity: a large funding deal undermined by the firm’s own investors.

“Whatever Waymo gains, Google Ventures loses,” said Stephen Diamond, associate professor of law at Santa Clara University.

An interesting dichotomy indeed.

Uber is buiding its own mapping solution

Leslie Hook for Financial Times:

Uber is preparing to pour $500m into an ambitious global mapping project as it seeks to wean itself off dependence on Google Maps and pave the way for driverless cars.

It is interesting to watch how car manufacturers (namely VAG, BMW, Daimler) and the bigger mobility companies are trying to cut their dependencies with Google Maps.

It is not surprising though: Google is very vocal about its Autonomous Driving ambitions and sooner than later will have to clarify how Google Maps fits in its overall strategy. Mapping will be strategic for the AV future and I do not think that Google will provide all the needed mapping features (e.g. accuracy) to direct competitors.

By developing its own maps Uber could eventually reduce its reliance on Google Maps, which currently power the Uber app in most of the world.

Although Google was an earlier investor in Uber, the two companies have avoided working closely together and are now developing rival technologies for driverless cars.

Last year Uber hired one of the world’s leading digital mapping experts, Brian McClendon, who previously ran Google Maps and helped create Google Earth.

“Accurate maps are at the heart of our service and backbone of our business,” Mr McClendon said in a statement. “The ongoing need for maps tailored to the Uber experience is why we’re doubling down on our investment in mapping.”

Competition is always good and brings nothing but better products!

Samsung invests in nuTonomy

Pulse News Korea reports: 

South Korea’s electronics giant Samsung Electronics Co. is stepping up its efforts to solidify its position in the automotive electronics industry through aggressive investment in smart car technology. 

According to multiple sources in the industry on Wednesday, Samsung Group’s unit Samsung Venture Investment Corp. invested in nuTonomy Inc., a venture that develops self-driving car software based on robotics, together with Signal Ventures Ltd. and Fontinalis Partners LLC and others. Following the successful fundraising campaign, the venture has raked in about $3.6 million (4.3 billion won) worth of funds in total, according to the industry sources. 

Cambridge, Massachusetts-based nuTonomy is a spin-off from the Massachusetts Institute of Technology (MIT), headed by Karl Iagnemma who has led the Robotic Mobility Group at MIT. With the investment, nuTonomy will focus on developing software that could enable an autonomous car to drive even in busy city traffic conditions, which is yet to be achieved by Google Inc. with its self-driving car project that proved to be a success on a highway. 

nuTonomy seems to be one of the few companies jointly with Google that is aiming to create an Operating System for autonomous vehicles. 

Most of the OEMs are focused on building those capabilities in house and consider the OS a central piece of their value proposition. My only question here is: would they be able to attract the right talent to build something that is well beyond their core business? 

Only time will tell…

Pearl, a new startup entering the aftermarket automated features for cars space

Alex Webb for Bloomberg:

Founded by a team of Apple veterans, Pearl Automation Inc. said Tuesday that its first product, a rear-view camera, will be available in September in the U.S. After the camera, which is similar to those that come standard on many new autos, the startup plans a slew of devices that can be built into your car to bring it up to speed with the latest driving capabilities.

There are new startups entering the AV and automated features for cars space every other week. I just can’t but wonder:

  1. For how long is it going to be profitable this market?
  2. How long before the market gets regulated and the party is over?

Also, the US has a long tradition of tinkering around with their cars.  However, this is far from “normal” in Europe beyond the tunning culture and the plug and play aftermarket gadgets. 

It is going to be interesting to watch how this market develops.