Machine Learning snapshot, June 2018

Kian Katanforoosh and Andrew Ng have been teaching CS230:Deep Learning, at Stanford. The project reports and posters list has just come out, summarizing work done by the students with help from the teaching team. More than 160 projects. It will be interesting to see which of these mature into applications.


Image from Painting Outside the Box: Image Outpainting with GANs (Mark Sabini, Gili Rusak) which was awarded first place in Outstanding Posters


Reports and posters

The 4th Research and Applied AI Summit just finished in London. The 125 slideset on the State of AI is a decent current snapshot of much more evolved work than the Stanford posters.

Moving things

More on Transportation as a System, prompted by an article in Ars Technica.


Autonomous vehicles for local delivery don’t need drivers; they can be smaller and lighter, slower, and at least as safe as US postal trucks.  The key to making this possible is to incent customers to collect their deliveries –  from the vehicle as it arrives, or from the nearest locker or pick-up point which can be reached by walking, perhaps with a small luggage trolley. In the US ‘rural’ post service puts letters in letter boxes which are on the street, not through your front door – but FedEx, UPS, and USPS leave packages on the doorstep.

Scheduling the delivery vehicle to arrive just when it’s convenient for the person collecting the packages to come down to the street from their multi-story building is the same kind of problem as arranging a Lyft pickup.

Solving this ‘last 50 feet’ issue of package delivery by getting the customer to fetch their packages for everything under certain sizes and weights will be much cheaper than trying to build robots to do the same job, and does not require an outside person or company to have access to the home (Amazon Key).

Amazon and Alibaba are likely to dominate global logistics; they have detailed knowledge about what their customers buy and can make supply chain predictions in order to get goods started on their journey prior to receiving specific orders. They are in a position to benefit hugely from building the overall integration of the systems for transporting goods.


Nuro self driving goods vehicle 

Adding AMA reddit thread by  Dave Ferguson from Nuro

Edge fabric management, Facebook version


Omar Baldinado introducing the Facebook Networking @ Scale event, 22 May 2018

The event was held at the Computer History Museum.


To pick out one talk for attention, Niky Riga described the Edge Fabric which Facebook uses to configure and operate its equipment in PoPs, of which it has many hundreds, generally located in shared co-location space near to its users, so as to be able to serve content with lower latency than would be possible for direct service from the large data centers. This was a refresh of the material from an paper published in ACM Sigcomm, available here

These events are good for meeting new people involved with networking, and for meeting up with people whom one hasn’t seen for a while, without the effort of going to Nanog or IETF meetings.

Updating, 22 June 18, to add the link to the video archive of the talks.


Artificial Intelligence for Institutional Investors


Venue – the National Press Club, 13th floor ballroom

The Markets Group runs events for institutional investors. I took part in a panel about Artificial Intelligence and Machine Learning, and in a round table discussion.

Diagrams and useful background sources listed below.

Definition of Artificial Intelligence, to illustrate that machine learning is a subset of AI. Sourced from the thorough review of AI in the NHS


Examples of AI and ML in use

At least 75% of the audience uses Netflix – which applies machine learning to improve its users’ experience of streaming as well as for content selection. Its results are driven by an extreme emphasis on keeping existing and attracting new customers. The data its customers generate by using it are used to make recommendations to them. Artwork personalization  Image discovery

View story at

Alibaba and Tencent Unlike the US companies, they have built their support for retailing primarily based on the customer’s mobile device; so they use facial recognition for identification, image scanning and matching for item selection, precise location specification in shopping venues and integration with payment apps to enhance the customer’s buying experience, both online and in person at a store. (joint report – Bain and Alibaba)

Recent reviews – longer reads  Recent developments are a direct result of the enormous improvement in computing capability at drastically reduced costs, a direct result of Moore’s Law – which continues.

Longer term effects of automation – 10 – 20 year horizon – Bain March 2018

13 artificial intelligence trends reshaping industries and economies – CBInsights February 2018

Capabilities at the end of 2017 – summary in slide format from Jeff Dean (Stanford, Google Brain team)


1950 Alan Turing asked “Can Machines Think”  Computing Machinery and Intelligence paper
1956 Initial definition of Artificial Intelligence at a workshop  at Dartmouth College   Remember the AI Winters in the 1970s, 1990s


Scaled Machine Learning


Bill Dally, Nvidia

Matroid and the Stanford center for image system engineering ran the 3rd year of the ScaledML conference yesterday, 24 March 2018. It was a concentrated survey of work in progress in machine learning, with admirably little overt advertising. The overall impression is of enormous actual potential, orthogonal to enormous hype and inflated expectations, significant uncertainty about what will actually get done, and of a lot of work in progress on necessary infrastructure in hardware, architecture, languages, systems, and education.

Agenda and speaker list

08:45 – 09:00 Introduction Reza Zadeh Matroid
09:00 – 10:00 Ion Stoica Databricks
10:00 – 11:00 Reza Zadeh Matroid
11:00 – 11:30 Andrej Karpathy Tesla
11:30 – 12:00 Jennifer Chayes Microsoft Research
13:00 – 14:00 Jeff Dean Google
14:00 – 14:30 Anima Anandkumar Amazon
14:30 – 15:00 Ilya Sutskever Open AI
15:00 – 15:30 Francois Chollet Google
16:00 – 17:00 Bill Dally Nvidia
17:00 – 17:30 Simon Knowles Graphcore
17:30 – 18:00 Yangqing Jia Facebook

From my notes :
The successor to the AMP Lab at Berkeley is RISE lab, building Real-time Intelligent Secure Explainable applications to make low-latency decisiongs on live data with strong security. (Ion Stoica). Note the remark about Explainable; this came up as a common theme.
Being able to examine detector errors and mistakes came up again in Reza Zadeh’s Matroid demonstration – this was the only live product shown. A user can build a detector with multiple attributes to pick out images from streaming video.
Bill Dally (Chief Scientist, Nvidia) reckons that Moore’s Law is dead; Simon Knowles (Graphcore) gave a more reasoned explanation about possible performance gains from hardware improvements over the next 10 years.


Graphcore hardware, use of BSP – Simon Knowles

Jeff Dean’s slides

Bill Dally’s slides

Anima Anandkumar

Ion Stoica

Francois Cholet on Keras

Ilya Sutskever

Jennifer Chayes

Yangqing Jia

Transporting people like IP packets


Transportation as a Service might be better thought of as transportation as a system; the items to be transported are people, and things.
An Economist special report, ‘Reinventing Wheels ‘, from early March 2018, discusses autonomous vehicles, urban planning, and possible changes in how people live, while ignoring a significant part of the overall transportation system ; there’s very little mention of pedestrians, and no mention of bicycles. This ignores the actual flexibility in the system if it allows for people to walk or bicycle for part of their journey. There are many urban and suburban trips where the time to drive and park a vehicle exceeds the time to walk or cycle the same trip.
Contrast this with ‘tim in Graz’ (Austria).
“At selected public transport stops, the tim locations, the Graz Lines bundle additional mobility services as a supplement to public transport:

  • e-car sharing
  • conventional car sharing
  • rental car for longer distances or long-term use
  • e-taxis with exclusive stand
  • public charging stations for private electric cars
  • bicycle parking

This offer makes it easier to dispense with your own car because it makes it easy and convenient to access a car when needed. You can also park your own e-car at the e-charging station and change to bus or train. Bicycle parking makes the change from the bike to public transport comfortable.”
Being able to change, at a location for which there is a business model that scales up, between different modes of transport is going to be important to the improvement of frequent, flexible, movement of people. Amazon, FedEx, DHL and the other shipping companies use this model for moving things.

TIM – Graz, Austria

Some people are usually early


Here’s something I’d like to be able to do. I’m usually early for things, and especially for flights, given the uncertainty of travel to airports, and the uncertainty of clearing security at airports. Some flight segments have very frequent service; I’d like to be able to get to the airport, clear security, then take the next flight with space available (whether space with legroom, or not) to where I’m going (which is maybe another airport for a connecting flight). I’d pay more for a ticket with that flexibility. Trying to get onto a different flight that the one you originally booked while you are at the airport usually requires going back outside security, and queuing for check-in along with all the people who have luggage to check; that’s not what I do.

This would require the ticket I booked originally to be treated like currency, so that it could be used as credit for the earlier seat; there should be a value to the airlines who can support this rebooking, because airline seats have a value up until the flight closes. If a flight leaves with empty seats for which someone would have paid, that’s a lost sale.

This app could start out by being be a representation of “same day flight changes” , kept up to date with all the policies and constraints described here.