iot and machine learning ciuf ajit jaokar

Upload: ajit-jaokar

Post on 03-Jun-2018

221 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    1/38

    Copyright : Futuretext Ltd. London0

    Slides for my talk on IoT and Machine LearningComputational Intelligence Unconference UK July 2014 -

    http://ciunconference.org/uk/2014/schedule.php@ajitjaokar

    [email protected]

    Sign up at www.futuretext.comto get copies of papers on IoT and

    Machine learning, Real time algorithms for IoT and Machinelearning algorithms for Smart cities

    http://ciunconference.org/uk/2014/schedule.phpmailto:[email protected]://www.futuretext.com/http://www.futuretext.com/mailto:[email protected]://ciunconference.org/uk/2014/schedule.php
  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    2/38

    Copyright : Futuretext Ltd. London1

    Ajit Jaokar

    -

    Machine Learning for IoT and Telecomsfuturetext applies machine learning techniques to complex problems in theIoT (Internet of Things) and Telecoms domains.

    We aim to provide a distinct competitive advantage to our customers throughapplication of machine learning techniques

    Philosophy:Think of NEST. NEST has no interface. Itsinterface is based on machine learning i.e. it

    learns and becomes better with use. This will becommon with ALL products and will determinethe competitive advantage of companies. Its awinner takes all game! Every product will have aself learning interface/component and the

    product which learns best will win!

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    3/38

    Copyright : Futuretext Ltd. London2

    Ajit Jaokar

    -

    [email protected]@futuretext.com

    http://www.futuretext.com/mailto:[email protected]:[email protected]://www.futuretext.com/
  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    4/38

    Copyright : Futuretext Ltd. London3

    Ajit Jaokar

    -

    Data is the new oil ...

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    5/38

    Copyright : Futuretext Ltd. London4

    Ajit Jaokar

    -

    The meek shall inherit the earth .. BUT not its mineral rights!Data is out there and is free (Open data). It provides no competitive advantages.Finding patterns in data is the holy grail (the oil!)

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    6/38

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    7/38

    Copyright : Futuretext Ltd. London6

    Ajit Jaokar

    -

    www.opengardensblog.futuretext.com

    World Economic ForumSpoken at MWC(5 times), CEBIT, CTIA, Web 2.0,CNN, BBC, Oxford Uni, Uni St Gallen, EuropeanParliament. @feynlabs teaching kids ComputerScience. Adivsory Connected Liverpool

    http://www.opengardensblog.futuretext.com/http://www.opengardensblog.futuretext.com/
  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    8/38

    Copyright : Futuretext Ltd. London7

    Image source: Guardian

    Image source: Guardian

    http://www.opengardensblog.futuretext.com/wp-content/uploads/2012/08/smart-cities-and-demand-for-data-scientists.jpghttp://www.opengardensblog.futuretext.com/wp-content/uploads/2012/08/smart-cities-and-demand-for-data-scientists.jpg
  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    9/38

    Copyright : Futuretext Ltd. London8

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    10/38

    Copyright : Futuretext Ltd. London9

    Ajit Jaokar

    IOT - THE INDUSTRY- STATE OF PLAY

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    11/38

    Copyright : Futuretext Ltd. London10

    Ajit Jaokar

    State of play - 2014 Our industry is exciting but mature - Now a two horse race fordevices with Samsung around 70% of Android Spectrum allocations and G cycles are predictable - 5G around 2020 50 billion connected devices by 2020 ITU world Radio communications Conference, November 2015. IOT has taken off .. not because of EU and Corp efforts but because ofMobile, kickstarter, health apps and iBeacon and ofcourse NEST(acquiredby Google)

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    12/38

    Copyright : Futuretext Ltd. London11

    Ajit Jaokar

    Stage One: Early innovation 1999 - 2007

    Regulatory innovationnet neutrality - Device innovation (Nokia7110 and Ericsson t68i) - Operator innovation (pricing, bundling,Enterprise) - Connectivity innovation (SMS, BBM)Content innovation (ringtones, games, EMS, MMS) - Ecosysteminnovation (iPhone)

    Stage two: Ecosystem innovation - iPhone and

    Android (20072010)

    Social innovation - Platform innovation - Communityinnovation - Long tail innovation - Applicationinnovation

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    13/38

    Copyright : Futuretext Ltd. London12

    Ajit Jaokar

    Phase three: Market consolidation2010 - 2013

    And then there were two ...Platform innovation and consolidationSecurity innovation

    App innovation

    Phase fourthree dimensions2014 ..Horizontal apps (iPhone and Android)Vertical (across the stack)hardware, security, DataNetwork5G and pricing

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    14/38

    Copyright : Futuretext Ltd. London13

    Ajit Jaokar

    Many of the consumer IOT cases will happen with iBeacon in the nexttwo years

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    15/38

    Copyright : Futuretext Ltd. London14

    Ajit Jaokar

    And 5G will provide the WAN connectivity 5G - Source Ericsson

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    16/38

    Copyright : Futuretext Ltd. London15

    Ajit Jaokar

    Samsung Gear Fit named Best Mobile Device of Mobile World

    Congress

    Notification or Quantification? Displays (LED, e-paper,Mirasol, OLED and LCD) - Touchscreen or hardware controls? -

    Battery life and charging

    http://androidandme.com/2014/03/news/samsung-gear-fit-named-best-mobile-device-of-mobile-world-congress/http://androidandme.com/2014/01/sponsored/whats-next-in-wearable-tech/
  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    17/38

    Copyright : Futuretext Ltd. London16

    Ajit Jaokar

    Hotspot 2.0

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    18/38

    Copyright : Futuretext Ltd. London17

    Ajit Jaokar

    Three parallel ecosystemsIOT is connecting things to the Internet which is not the same asconnecting things to the cellular network!The difference is money .. and customers realise it

    IOT local/personal(iBeacon, Kickstarter, Health apps)

    M2MMachine to Machine

    IOT pervasive(5G, Hotspot 2.0)

    Perspectives 2014 2015(radio conf) 2020(5G, 2020) 2014 iBeacon (motivate retailers to open WiFi) Hotspot 2.0 connect cellular and wifi worlds

    Default wifi and local world? Operator world (Big)Data, Corporate, pervasive apps really happen

    beyond 2020 So 5G will be timed well. The ecosystems will develop and they will be

    connected by 5G

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    19/38

    Copyright : Futuretext Ltd. London18

    Ajit Jaokar

    IOT INTERNET OF THINGS

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    20/38

    Copyright : Futuretext Ltd. London19

    As the term Internet of Things implies (IOT) IOT is about Smart

    objectsFor an object (say a chair) to be smart it must have three things- An Identity (to be uniquely identifiable via iPv6)- A communication mechanism(i.e. a radio) and- A set of sensors / actuators

    For example

    the chair may have a pressure sensor indicating that it is occupiedNow, if it is able to know who is sitting it could co-relate more data byconnecting to the persons profileIf it is in a cafe, whole new data sets can be co-related (about the venue,about who else is there etc)

    Thus, IOT is all about Data ..

    IoT != M2M (M2M is a subset of IoT)

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    21/38

    Copyright : Futuretext Ltd. London20

    Sensors lead to a LOT of Data (relative to mobile) .. (source David

    wood blog)

    By 2020, we are expected to have 50 billion connected devicesTo put in context:The first commercial citywide cellular network was launched in Japan by NTTin 1979The milestone of 1 billionmobile phone connections was reached in 2002

    The 2 billionmobile phone connections milestone was reached in 2005

    The 3 billionmobile phone connections milestone was reached in 2007

    The 4 billionmobile phone connections milestone was reached in February

    2009.

    Gartner: IoT will unearth more than $1.9 trillionin revenue before 2020; Cisco thinksthere will be upwards of50 billion connected devicesby the same date; IDC estimatestechnology and services revenue will grow worldwide to$7.3 trillion by 2017(upfrom$4.8 trillion in 2012).

    http://www.gartner.com/newsroom/id/2636073http://share.cisco.com/internet-of-things.htmlhttp://www.idc.com/getdoc.jsp?containerId=prUS24671614http://www.idc.com/getdoc.jsp?containerId=prUS24366813http://www.idc.com/getdoc.jsp?containerId=prUS24366813http://www.idc.com/getdoc.jsp?containerId=prUS24671614http://share.cisco.com/internet-of-things.htmlhttp://www.gartner.com/newsroom/id/2636073
  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    22/38

    Copyright : Futuretext Ltd. London21

    So, 50 billion by 2020 is a large number

    Smart cities can be seen as an application domain of IOT

    In 2008, for the first time in history, more than half of the worldspopulation will be living in towns and cities.By 2030 this number will swell to almost 5 billion, with urban growthconcentrated in Africa and Asia with many mega-cities(10 million +

    inhabitants).By 2050, 70% of humanity will live in cities.

    Thats a profound change and will lead to a different management approachthan what is possible todayAlso, economic wealth of a nation could be seen as Energy +Entrepreneurship + Connectivity (sensor level + network level +

    application level)Hence, if IOT is seen as a part of a network, then it is a core component ofGDP.

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    23/38

    Copyright : Futuretext Ltd. London22

    Ajit Jaokar

    Machine Learning

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    24/38

    Copyright : Futuretext Ltd. London23

    What is Machine Learning?

    Mitchell's Machine LearningTom Mitchell in his book Machine Learning Thefield of machine learning is concerned with the question of how to construct computerprograms that automatically improve with experience.

    formally:A computer program is said to learn from experience E with respect to

    some class of tasks T and performance measure P, if its performance attasks in T, as measured by P, improves with experience E.

    Think of it as a design tool where we need to understand:What data to collect for the experience (E)What decisions the software needs to make (T) andHow we will evaluate its results (P).

    A programmers perspective:Machine Learning involves:a) Training of a model from datab) Predicts/ Extrapolates a decisionc) Against a performance measure.

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    25/38

    Copyright : Futuretext Ltd. London24

    What Problems Can Machine Learning Address? (source JasonBrownlee)

    Spam Detection: Credit Card Fraud Detection Digit Recognition: Speech Understanding: Face Detection: Product Recommendation:

    Medical Diagnosis: Stock Trading: Customer Segmentation

    Shape Detection.

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    26/38

    Copyright : Futuretext Ltd. London25

    Types of Problems

    Classification: Data is labelled meaning it is assigned a class, for examplespam/nonspam or fraud/nonfraud. The decision being modelled is toassign labels to new unlabelled pieces of data. This can be thought of as adiscrimination problem, modelling the differences or similarities between groups.

    Regression:Data is labelled with a real value ratherthan a label. Examples that are easy to understand are time series data like the price ofa stock over time. The decision being modelled is the relationships betweeninputs and outputs.

    Clustering: Data is not labelled, but can be divided into groups based onsimilarity and other measures of natural structure in the data.

    An example from the above list would be organising pictures by faces without

    names, where the human user has to assign names to groups, like iPhoto on the Mac.

    Rule Extraction:Data is used as the basis for the extraction ofpropositional rules (antecedent/consequent or ifthen).Often necessary to work backwards from a Problem to the algorithm and then work withData. Hence, you need a depth of domain experience and also algorithm experience

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    27/38

    Copyright : Futuretext Ltd. London26

    What Algorithms Does Machine Learning Provide?

    RegressionInstance-based Methods

    Decision Tree Learning

    Bayesian

    Kernel Methods

    Clustering methods

    Association Rule LearningArtificial Neural Networks

    Deep Learning

    Dimensionality Reduction

    Ensemble Methods

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    28/38

    Copyright : Futuretext Ltd. London27

    Ajit Jaokar

    IoT and Machine Learning

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    29/38

    Copyright : Futuretext Ltd. London

    Basic idea of machine learningis to build a mathematical model based on

    training data(learning stage) predict results for new data(predictionstage) and tweak the model based on new conditions

    What type of model? Predicitive, Classification, Clustering, Decision

    Oriented, Associative

    IoT and Machine Learning

    On one hand - IoT creates a lot of contextual data which complements existingprocesses

    On the other handthe Sheer scale of IoT calls for unique solutions

    Types of problems:

    Apply existing Machine Learning algorithms to IoT data

    Use IoT data to complement existing processes Use the scale of IoT data to gain new insights

    Consider some unique characteristics of IoT data (ex streaming)

    28

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    30/38

    Copyright : Futuretext Ltd. London29

    IoT : from traditional computing to ..

    Gone from making Smart things smarter(traditional computing) to

    a) Making dumb things smarter .. andb) living things more robust

    3 Domains:

    Consumer, Enterprise, Public infrastructure

    1) Consumerbio sensors(real time tracking), Quantified selffocussing onbenefits

    2) EnterpriseComplex machinery (preventative maintenance), asset efficiencyreducing assets, increasing efficiency of existing assets. More from transactions torelationships(real time context awareness).

    3) Public infrastructure(Dynamically adjust traffic lights). Dis-economies ofscale(bad things also scale in cities)Thanks John Hagel III

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    31/38

    Copyright : Futuretext Ltd. London30

    Three key areas:

    a) Move from exception handling to patterns of exceptions over time.(aresome exceptions occurring repeatedly? Do I need to redsign my product, Is that anew product?)

    b) Move from optimization to disruptionownership to rental ship (Where are allthese dynamic assets?)

    c) Move to self learning: Robotics: From assembly line to self learning

    robots(Boston Dynamics), autonomous helicopters

    Four examples of differences:Sensor fusion - Deep Learning - Real time - Streaming

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    32/38

    Copyright : Futuretext Ltd. London

    Sensor fusion

    Sensor fusionis the combining of sensory data or data derived from

    sensory data from disparate sources such that the resulting informationis in some sense better than would be possible when these sources were

    used individually. The data sources for a fusion process are not specified

    to originate from identical sensors. Sensor fusion is a term that covers a

    number of methods and algorithms, including: Central Limit Theorem,

    Kalman filter, Bayesian networks, Dempster-Shafer

    Example: http://www.camgian.com/http://www.egburt.com/

    31

    http://www.camgian.com/http://www.egburt.com/http://www.egburt.com/http://www.egburt.com/http://www.camgian.com/http://www.camgian.com/http://www.camgian.com/
  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    33/38

    Copyright : Futuretext Ltd. London

    Deep learning

    Google's acquisition ofDeepMind Technologies

    In 2011, Stanford computer science professor Andrew Ng foundedGoogles Google Brain project, which created a neural network trainedwith deep learning algorithms, which famously proved capableofrecognizing high level concepts, such as cats, after watching justYouTube videos--and without ever having been told what a cat is.

    A smart-object recognition algorithm that doesnt need humanshttp://www.kurzweilai.net/a-smart-object-recognition-algorithm-that-doesnt-need-humansA feature construction method for general object recognition (Kirt Lillywhite,Dah-JyeLee n, BeauTippetts, JamesArchibald)

    32

    http://deepmind.com/http://www.nytimes.com/2012/06/26/technology/in-a-big-network-of-computers-evidence-of-machine-learning.html?_r=0http://www.kurzweilai.net/a-smart-object-recognition-algorithm-that-doesnt-need-humanshttp://www.kurzweilai.net/a-smart-object-recognition-algorithm-that-doesnt-need-humanshttp://www.kurzweilai.net/a-smart-object-recognition-algorithm-that-doesnt-need-humanshttp://www.kurzweilai.net/a-smart-object-recognition-algorithm-that-doesnt-need-humanshttp://www.kurzweilai.net/a-smart-object-recognition-algorithm-that-doesnt-need-humanshttp://www.kurzweilai.net/a-smart-object-recognition-algorithm-that-doesnt-need-humanshttp://www.kurzweilai.net/a-smart-object-recognition-algorithm-that-doesnt-need-humanshttp://www.kurzweilai.net/a-smart-object-recognition-algorithm-that-doesnt-need-humanshttp://www.kurzweilai.net/a-smart-object-recognition-algorithm-that-doesnt-need-humanshttp://www.kurzweilai.net/a-smart-object-recognition-algorithm-that-doesnt-need-humanshttp://www.kurzweilai.net/a-smart-object-recognition-algorithm-that-doesnt-need-humanshttp://www.kurzweilai.net/a-smart-object-recognition-algorithm-that-doesnt-need-humanshttp://www.kurzweilai.net/a-smart-object-recognition-algorithm-that-doesnt-need-humanshttp://www.kurzweilai.net/a-smart-object-recognition-algorithm-that-doesnt-need-humanshttp://www.kurzweilai.net/a-smart-object-recognition-algorithm-that-doesnt-need-humanshttp://www.kurzweilai.net/a-smart-object-recognition-algorithm-that-doesnt-need-humanshttp://www.kurzweilai.net/a-smart-object-recognition-algorithm-that-doesnt-need-humanshttp://www.kurzweilai.net/a-smart-object-recognition-algorithm-that-doesnt-need-humanshttp://www.kurzweilai.net/a-smart-object-recognition-algorithm-that-doesnt-need-humanshttp://www.kurzweilai.net/a-smart-object-recognition-algorithm-that-doesnt-need-humanshttp://www.kurzweilai.net/a-smart-object-recognition-algorithm-that-doesnt-need-humanshttp://www.nytimes.com/2012/06/26/technology/in-a-big-network-of-computers-evidence-of-machine-learning.html?_r=0http://deepmind.com/
  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    34/38

    Copyright : Futuretext Ltd. London33

    Real time:Beyond Hadoop (non hadoopable) the BDAS stack

    BDAS Berkeley data analytics stack

    Spark an open source, in-memory, cluster computing framework.Integrated with Hadoop(can work with files stored in HDFS)Written in Scala

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    35/38

    Copyright : Futuretext Ltd. London34

    Real time (Stream processing)

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    36/38

    Copyright : Futuretext Ltd. London35

  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    37/38

    Copyright : Futuretext Ltd. London36

    Ajit Jaokar

    -

    [email protected]@futuretext.com

    http://www.futuretext.com/mailto:[email protected]:[email protected]://www.futuretext.com/
  • 8/12/2019 IoT and Machine Learning CIUF Ajit Jaokar

    38/38

    37

    @[email protected]

    Sign up at www.futuretext.comto get copies of papers on IoT andMachine learning, Real time algorithms for IoT and Machine

    learning algorithms for Smart cities

    mailto:[email protected]://www.futuretext.com/http://www.futuretext.com/mailto:[email protected]