The $84 Billion AdTech Industry, Single-Machine AI and the Mbed IoT OS (#wic)
This week on the Argus Labs blog we look at some impressive AdTech numbers – 150 milliseconds RTB, the 60-80% AdTech ‘technology tax’, .. – and hear why you do not have to be Google to use deep learning techniques to build an artificial brain. There is also wearables and wellness, a free battery-friendly OS for the Internet of Things and thoughts machine consciousness.
This Week in Context
Your Weekly Update on All Things Context, October 3 2014
Happy to announce we are nominated as Rising Star for the Deloitte Tech Fast50. This means that if you can’t make it to Ad:Tech London to hear our Founder our founder talk on “Context-aware mobile advertising: Turning ads into real-time engagement”, there is an excellent opportunity to meet up in Brussels on October 22. Propose by tweet: @sentiance.
“Training a deep neural net is still just as much an art as a science. Many parameters used to train neural networks are based on intuition.”
– Kaggle chief scientist Ben Hamner, wired.com
1. You Don’t Have to Be Google to Build an Artificial Brain
It’s all a question of size: the size of your database and how fast a turnaround you want, but you don’t need 16,000 machines to harness the power of ‘deep learning’, explains Richard Socher in an interview with Wired. Using AI technology requires expertise, but most data scientists use single machines, and the parallel processing capabilities of their GPU cards, to solve their problems using deep learning algorithms.
“There are only so many companies have datasets the size of Google’s and Facebook’s and Yahoo’s,” says Socher, who only used single machines in his own deep learning work. “Other, normal companies have smaller datasets, and they can train models too.”
2. Little Brother, or How Technology is Changing the Ad Business
The Economist has a special report on advertising and technology, covering everything from the ‘information war’ between data brokers, sellable segments and programmatic bidding, to privacy and the idea of a “fair data” movement. Some impressive numbers in the series as well:
- Online advertising economy: $120 billion (of which $19 billion for mobile ad)
- BlueKai holds about 1 billion profiles (average 50 attributes each)
- 50 to 70 data points about users on desktops, 100 on mobile
- Real-time bidding: 150 milliseconds from first page load to ad display
- 20% of online adverts sold using RTB
- “Technology tax” of 60-80% that goes to adtech firms, rather than the publisher (70% of $120 billion = $84 billion)
- Kellogg’s has a price a price benchmark of $5.40 per 1,000 impressions
- Personal data privacy remains a concern for around 75% of consumers in most countries
- Google and Facebook alone control over 47% of all digital advertising in America, 57% of mobile
To demonstrate it is not just “dodgy” websites that are flooded with trackers, here is the Ghostery overview of the trackers running on the Economist’s “Moving Targets” article. There are over thirty, including the above mentioned BlueKai:
3. Could wearable tech help us to improve our mental health?
Becca Caddy covers the future of wearables and wellness, as she asks if wearable tech will allow us to better understand our moods, emotions and mental health problems. While wearables are a help to mitigate the issue that health professionals can’t constantly monitor their patients, one of the challenges that remains is using the data to make truly meaningful changes. (Suggestion by Koen)
4. What It Will Take for Computers to Be Conscious
Christof Koch, interviewed by the MIT Technology Review, on how a software model of the brain could never be conscious, but machines themselves could:
“I think consciousness, like mass, is a fundamental property of the universe. The analogy, and it’s a very good one, is that you can make pretty good weather predictions these days. You can predict the inside of a storm. But it’s never wet inside the computer. You can simulate a black hole in a computer, but space-time will not be bent. Simulating something is not the real thing.”
5. Mbed: a free, customisable operating system for the Internet of Things
Chip designer ARM officially announced Mbed, a free operating system for the Internet of Things optimised to consume low amounts of power. ARM hopes Mbed will be used in everything from streetlights to home appliances, and even in simple wearables that predominantly collect readings, such as heart rate. Parts of the Mbed operating system will be open source, other will remain under control of ARM, in a bid to prevent fragmentation.
A Work of Art & Tech
The Creators Project highlights the work of Marion Balac, who created her work around one of algorithmic technology’s faults: having a hard time taking the context of a face into account, and making no distinction between a human face and its reproduction on — let’s say a public statue or a Sphinx.
“By applying this bureaucratic tool to figures of faith, plunging them into anonymity, we can really see the robot’s point of view,” Balac told The Creators Project. “It treats every human face it encounters as data, without exceptions, neglecting religious or scale principles.”
(The reverse – cases where Google forgot to remove ‘itself’, its own presence, from the picture – is the subject of the art project GoogleGeist.)
Papers, Talks & Research
- Reading Emotions Through Affective Computing: Rosalind Picard (affective computing, media, video)
- Final recommendations in relation to privacy in the convergent society: a user empowerment perspective (privacy, social networks, data protection, English research report and Dutch summary)
- Privacy in a World of Mobile Devices (privacy, ubiquitous computing, IoT, personal privacy policies, paper)
- The new profiling: Algorithms, black boxes, and the failure of anti-discriminatory safeguards in the European Union (security, profiling, automated decisions, paper)
- Regulating Emerging Robotic Technologies in Europe: Robotics facing Law and Ethics Funding scheme: Collaborative project (RoboLaw, ethics, legislation, report and The Economist article on robot jurisprudence)
- Human Brain Project Achievements: Year One (artificial brain, ethics, HBP Year One Report)
Enjoy the reads, and have a great weekend!
And if you haven’t done so yet, kindly consider subscribing to the Week in Context here.