CHAPTER 4: TECHNOLOGY
fuel for large data sets [Big Data].44 The development of tools to aid visualisation of
Big Data is a growth industry too. It is predicted that there will be 28 billion IOT devices
by 2020,45 and the data transmission speeds made possible by the next generation of
mobile network (5G) will fuel this growth.
4.37.
Furthermore, IOT is expected to increase the use of cloud computing services: indeed
it is predicted that in the next five years 90% of IOT data will be hosted via cloud
services. Cloud computing is the term used to describe the delivery of computing
resources over the internet on demand. Users can access software via the cloud
rather than purchase the software. Another aspect of cloud services is the storing and
accessing of data. This makes cloud computing an ideal storage system for IOT as it
provides the ability to respond quickly to changes in demand and supply. Since the
beginning of 2015, two telecommunications companies have launched cloud-based
products to handle data generated by IOT.46
Machine learning technologies
4.38.
Growth in computer processing capacity and data sets has led to advances in a branch
of artificial intelligence called Deep Learning.47 Deep Learning software mimics the
structure of the human brain in order to train computers to see patterns. Research
published at the end of 2014 described how image-recognition software is now capable
of recognizing and describing scenes, rather than just identifying objects in scenes.
The software was developed by training computers to see patterns in pictures and their
description using neural networks.48
4.39.
The Biometrics Commissioner has highlighted the fact that there have been substantial
developments in both automated facial and speaker recognition systems in the last
few years.49 The technique involved in Deep Learning is at the heart of some of these
recent developments in biometric systems. It has been applied in the area of facial
recognition to develop software called Deep Dense, which is able to determine whether
an image contains a face, even if part of the face is hidden or upside down. 50 Open
Rights Group’s submission to the Review highlighted that machine learning technology
has been used to teach computers to classify faces based on attributes such as facial
expression or hair style. It is also behind advances in speaker recognition systems.
The NSA Technology Transfer programme 2013/2014 lists an invention capable of
real-time simultaneous identification of multiple voices. One of three future trends in
44
45
46
47
48
49
50
See 8.65 onwards for the use of Big Data by private companies. Examples of how Big Data can be
used for the common good can be found at http://www.nesta.org.uk/publications/data-good.
This was the figure quoted by IBM from analyst firm IDC in announcing cloud services for IOT devices:
see
http://www.theinquirer.net/inquirer/news/2376409/ibm-announces-internet-of-thingscloudservices.
Blackberry announced this on its website: http://press.blackberry.com/press/2015/blackberry-unveilscloud-based-internet-of-things-platform-.html, and AT&T’s launch was announced in early January
2015:
http://www.computerworld.com/article/2864069/att-builds-on-internet-of-things-offerings-withcloud-based-data-store.html.
As set out in some detail in MIT Technology Review, 10 Breakthrough Technologies 2013, see
http://www.technologyreview.com/featuredstory/513696/deep-learning/
IBW Watson uses Deep
Learning techniques.
See e.g. J. Markoff, “Researchers Announce Advance in Image Recognition Software”, NY Times, 17
November 2014.
Biometrics Commissioner: Annual Report 2013-2014, para 336.
“”Deep Dense Face Detector” a breakthrough in face detection”, TechWorm website, 20 February 2015.
58