00:00:01
[Music]
00:00:10
[Music]
00:00:29
dear participants in today's session we
00:00:31
shall discuss about the applications of
00:00:33
artificial intelligence in Internet of
00:00:36
things as we know internet is
00:00:38
transforming from the internet of
00:00:40
computers to the internet of things over
00:00:42
the years massively interconnected
00:00:45
systems which are also known as cyber
00:00:48
physical systems are emerging from the
00:00:50
assimilation of many facets like
00:00:53
infrastructure embedded devices smart
00:00:55
objects humans and physical
00:00:58
environment this all is is heading to a
00:01:00
huge internet of everything in a smart
00:01:02
cyber physical Earth uh internet of
00:01:06
things and cyber physical systems when
00:01:08
they are conjugated with data science it
00:01:11
may emerge as the next smart Revolution
00:01:14
thus iot with AI can become a huge
00:01:17
breakthrough easing human life however
00:01:20
there are certain concerns also in this
00:01:23
which include security concerns and
00:01:25
ethical
00:01:26
issues uh another issue which is
00:01:29
impeding the growth of this field is a
00:01:31
huge data is generated and weak
00:01:33
computation power is available as of now
00:01:35
so all these are areas of study to make
00:01:38
this more uh properly applicable in the
00:01:42
uh real world as we know internet of
00:01:45
things is something which anticipates a
00:01:48
world saturated with intelligent gadgets
00:01:50
which are frequently called smart
00:01:52
objects and these smart objects are
00:01:54
interconnected through internet or any
00:01:56
other communication medium with the
00:01:59
Advent of internet of everything in a
00:02:01
smart cyber physical Earth uh we have
00:02:04
seen that certain concerns like huge
00:02:06
data which has been generated with much
00:02:08
weaker computation power uh is something
00:02:12
which is impeding its growth the answer
00:02:14
to this is research in Ai and data
00:02:16
science because through data science we
00:02:19
can find uh the answers to how to manage
00:02:22
this huge amount of data thus iot and AI
00:02:25
can become a huge bre breakthrough once
00:02:27
they're combined together when we talk
00:02:30
of smartness Big Data analysis machine
00:02:34
learning is the way to
00:02:36
go uh human intelligence basically is
00:02:39
taking a perfect decision at an
00:02:41
appropriate time when we talk of
00:02:43
artificial intelligence it is a take
00:02:45
away from Human intelligence so
00:02:48
artificial intelligence would be
00:02:49
choosing a right decision at the
00:02:51
appropriate
00:02:52
time when we discuss AI it is not a
00:02:56
standalone discipline it is a has a
00:02:58
interdisciplinary nature
00:03:00
Fields as diverse as philosophy computer
00:03:03
science mathematics statistics biology
00:03:07
physics sociology psychology Etc all
00:03:09
combined together in AI so AI if you are
00:03:15
defining it in a formal way is the
00:03:16
science of installing intelligence in
00:03:18
machines so that they are capable of
00:03:21
doing tasks that are traditionally
00:03:23
required the human mind so whatever we
00:03:26
are doing through our mind we are trying
00:03:28
to simulate or we are trying to emulate
00:03:30
it through AI data is something which is
00:03:34
uh generated while doing all this and in
00:03:36
real world data has some very unwelcome
00:03:38
properties that includes huge volume of
00:03:41
data and that huge volume is also
00:03:43
unstructured in nature the data sources
00:03:46
are varied and it needs real time
00:03:50
processing also it changes
00:03:52
continuously so what is the way forward
00:03:55
machine learning is the way forward and
00:03:57
machine learning basically gives inan
00:04:00
systems the ability to learn without
00:04:02
actually having to program them
00:04:04
explicitly it is about learning from
00:04:06
experience rather than uh actually
00:04:09
programming for its functioning so
00:04:11
adaptive learning is what machine
00:04:13
learning is all about so many uh
00:04:16
adaptive learning algorithms have used
00:04:20
uh what is happening in various uh uh
00:04:24
devices or various things in the real
00:04:26
world or various uh species in the real
00:04:30
world for example earthworm is a uh
00:04:34
inspiration for the development of smart
00:04:36
thermostats and adaptive learning of new
00:04:39
responses is the quality which is uh
00:04:42
actually picked up from the earth form
00:04:43
to develop a smart thermostat similarly
00:04:46
learning by trial and error which is
00:04:48
picked up from fish that was used to
00:04:51
develop a Kronos Robo then learning by
00:04:54
setting a goal acting to achieve it and
00:04:56
then assessing itself this quality or
00:04:59
this this property of the octopus was
00:05:01
used to develop the application in COG
00:05:05
self-consciousness and higher order
00:05:07
thinking which is something which
00:05:09
chimpanzees exhibit that is used in
00:05:12
development of Siri which we all use in
00:05:14
our
00:05:15
smartphones uh emotions like frustration
00:05:18
and happiness which a small child from 1
00:05:20
to 6 year old uh those emotions are used
00:05:24
to develop Cosmo so that is a object
00:05:28
which is developed basically simulating
00:05:30
the emotions of a 1 to six-year-old
00:05:32
child now full theory of Mind interpret
00:05:35
human emotions and Responding back
00:05:38
accordingly which is something a 7 to 11
00:05:41
year old child does it is programmed
00:05:44
into a device called a robo called
00:05:48
pepper then during test simulation and
00:05:51
experiments used a 12 plus year old
00:05:53
human mind and this particular thing
00:05:56
this test was devised by mit's AI
00:05:59
program by Eugene
00:06:01
ghostman now when we talk of Internet of
00:06:04
things so many fields are merging in it
00:06:06
into it for example high performance
00:06:08
Computing cognitive science realtime
00:06:11
analytics big data analytics machine
00:06:14
learning wireless sensor networks then
00:06:17
Network Control software actuators
00:06:20
sensors embedded systems they are all
00:06:22
part of or they all uh merge into the
00:06:25
field of iot so when we talk of all
00:06:28
these things like iot intelligence and
00:06:31
like that there are certain relations
00:06:33
which we can Define to actually
00:06:37
understand how they are interl how they
00:06:39
merge into each other things when they
00:06:42
are intersected with intelligence we get
00:06:45
what is called smart objects or smart
00:06:46
devices similarly a network intersected
00:06:50
with intelligence will give you a smart
00:06:52
Network then things intersected with net
00:06:55
Network would give us network
00:06:58
devices similarly services and
00:07:01
intelligence once they intersected we
00:07:03
get smart Services services and network
00:07:06
intersected gives us internet services
00:07:09
then things and intelligence and network
00:07:13
the intersection of all these things
00:07:15
will give us internet of things internet
00:07:18
services and intelligence once they are
00:07:20
intersected their common cusp gives us
00:07:23
internet of
00:07:25
services and then internet of things
00:07:28
it's Union with internet of services
00:07:30
will give us internet of everything
00:07:32
which would be having physical as well
00:07:34
as smart devices so things and
00:07:37
everything when we talk of these two
00:07:39
terms when we are talking about iot and
00:07:41
ie we must be very clear about the
00:07:44
concept of things and the concept of
00:07:48
everything one straightforward concept
00:07:50
that may come to the mind is anything
00:07:52
that can be connected maybe the thing in
00:07:55
iot however we we can Define it in the
00:07:58
other way around also
00:07:59
there can be more features in making a
00:08:01
physical object a thing the thing which
00:08:04
could be a living thing or a non-living
00:08:06
thing should have a way to generate or
00:08:09
collect data that is the first property
00:08:10
a thing should have a way to process
00:08:13
data a way to spend or receive data a
00:08:17
way to identify itself so thing is a
00:08:20
physical object whereas IO is the things
00:08:24
and intelligent
00:08:26
Services Services as we know are uh not
00:08:30
uh physical they are basically abstract
00:08:34
uh when we talk of AI enabled iot there
00:08:36
are some examples of the existing iot
00:08:38
services with the working of AI behind
00:08:40
them and these examples we use in our
00:08:43
everyday life also for example we use
00:08:46
voice assistants like Alexa Siri and so
00:08:49
on in our smartphones then certain Robos
00:08:52
like pepper sopia they are also be in
00:08:55
application in use in different
00:08:57
applications then robotic kitchens are
00:08:59
there there which are actually being
00:09:01
used smart devices like smart ens Sky
00:09:05
smart lights automative Ai and certain
00:09:08
industrial iot applications like primer
00:09:11
and Pluto shift they are all part of our
00:09:13
AI enabled iot uh applications which are
00:09:16
presently deployed in the real world so
00:09:20
once we are talking about AI enabled iot
00:09:22
we should understand that what cyber
00:09:24
physical systems are in a formal way
00:09:27
this term cyber physical systems it
00:09:29
emerged around 2006 and it was coined by
00:09:33
Helen Gil at the National Science
00:09:34
Foundation in the United
00:09:36
States uh as for as far as NSF or the
00:09:40
National Science Foundation definition
00:09:42
is concerned they say it is an
00:09:44
engineered system that are built from
00:09:47
and depend upon the seamless integration
00:09:49
of computational algorithms and physical
00:09:52
components for example let us consider
00:09:55
an example of what we can visualize in
00:09:57
terms of cyber physical systems
00:09:59
you are going in a car somewhere the car
00:10:02
meets with an
00:10:03
accident normally what is happening is
00:10:06
once it meets with an accident accident
00:10:08
somebody calls an ambulance somebody
00:10:10
would call police you will reach an
00:10:12
Hospital in that hospital the doctors
00:10:15
would ask that okay this is a police
00:10:17
case so let the police register a case
00:10:20
first uh this is something which would
00:10:22
waste very precious time in a accident
00:10:26
case uh what would happen if you
00:10:29
visualize a system in which once that
00:10:30
accident happens the system is
00:10:33
immediately informing the nearest uh
00:10:36
hospital and the system is informing the
00:10:38
nearest police station and the police
00:10:40
reaches before or parall to when you
00:10:42
reach the hospital in an ambulance so
00:10:45
that way a lot of time is saved so that
00:10:48
is what we are visualizing in a CPS kind
00:10:50
of a system CPS again like uh iot is a
00:10:54
combination of so many disciplines and
00:10:56
this machine learning which is Big use
00:10:59
use in this for example machine learning
00:11:01
is basically it is a platform to learn
00:11:03
the trends of the system from data
00:11:05
generated in the past to make an
00:11:07
informed decision in future without
00:11:09
manual
00:11:10
monitoring then big data analytics or
00:11:12
what we call data science that is
00:11:15
another uh area which machine learning
00:11:18
uh uses so all the data that is
00:11:20
generated in a huge interconnected
00:11:22
system will constitute a massive content
00:11:25
that will be processed and analyzed over
00:11:27
time to make the system better
00:11:30
so usually machine learning algorithms
00:11:32
are modified and adapted to handle this
00:11:34
big data scenario then design is another
00:11:38
area which is part of machine learning
00:11:40
the overall system needs a robust
00:11:42
tolerant and efficient design which
00:11:44
connects all the components as required
00:11:47
process science its usage would be
00:11:50
different industrial manufacturing
00:11:51
processes are demanding the use of
00:11:53
Automation in their production lines
00:11:56
then wireless sensor networks for
00:11:58
communication are required the whole
00:12:00
system would depend on communication
00:12:02
wireless connections between each
00:12:04
component of the system would help pass
00:12:06
information from one object to the
00:12:09
another So within a system or within
00:12:11
objects uh that would be required
00:12:14
software is the heart of all this
00:12:17
functioning all the working devices in
00:12:19
system need a software to work these
00:12:21
softwares would be system and task
00:12:24
specific then embedded systems the
00:12:27
gadgets the devices that constitute a
00:12:29
CPS would contain uh embedded systems
00:12:32
like camera temperature measurement
00:12:35
sensors each device would have different
00:12:37
embedded systems or sensors as per their
00:12:40
requirements cybernetics is another
00:12:42
field of study which is relevant to
00:12:45
Mechanical physical biological cognitive
00:12:48
and social systems uh mechatronics and
00:12:51
Robotics these are fields which seek
00:12:53
humanlike actions for different tasks so
00:12:56
these will not be manually handled for
00:12:57
giving given some fixed instructions
00:13:00
rather they will be intelligent enough
00:13:02
to know what needs to be done at the
00:13:04
right time then high performance or
00:13:06
cloud computing is something which is
00:13:08
part of the machine Learning Systems
00:13:11
typically the issues considered cannot
00:13:13
be solved on a single commodity computer
00:13:16
because of the uh huge amount of data So
00:13:19
within a sensible amount of time you
00:13:21
can't handle all that so excessively
00:13:23
complex operations are required or the
00:13:26
execution is incomprehensible because of
00:13:29
the restricted accessible assets a lot
00:13:32
of training data is required so these
00:13:35
systems are required here then cognitive
00:13:37
science what is cognitive science it is
00:13:40
predominantly involving Concepts from
00:13:42
psychology philosophy neuros
00:13:45
Neuroscience then anthropology computer
00:13:47
science and Linguistics so it is a study
00:13:50
of mind and the level of intelligence
00:13:52
the goal is to understand the nature of
00:13:55
knowledge in various living beings and
00:13:56
how that knowledge is acquired processed
00:13:59
and used so that we can then mimic in
00:14:01
our
00:14:03
systems so what are the various
00:14:05
components in iot CPS system when we
00:14:08
talk of its components the iot
00:14:10
architecture tree would have internet of
00:14:13
things and that you'll have things
00:14:15
things can be a smart object or a
00:14:19
identity Network then data storage and
00:14:22
data processing which we call data
00:14:24
science then security is another
00:14:26
component so what a smart SM objects
00:14:29
actually we've been talking about this
00:14:30
term two elements to consider in the
00:14:33
physical world are a physical entity and
00:14:35
a smart object a physical entity or a
00:14:39
physical thing can be represented in the
00:14:41
cyber world by a d by its digital proxy
00:14:45
which is called a DP so digital
00:14:47
properties have two fundamental
00:14:49
properties one is each digital proxy
00:14:52
must have a single ID that distinguishes
00:14:55
it from others the association between
00:14:57
the digital proxy and the physical
00:14:59
entity must be established automatically
00:15:01
through this
00:15:04
DP then relevant digital parameters
00:15:07
pertaining to the characteristics of the
00:15:09
physical entity can be refreshed upon
00:15:10
any difference in the former similarly
00:15:13
changes that influence the digital proxy
00:15:16
might be shown on the physical entity in
00:15:18
the physical world through actuators so
00:15:20
that is how we can show it in the
00:15:23
physical world we can use actuators for
00:15:25
that then data storage and data
00:15:27
processing it that is the main motive of
00:15:30
iot and CPS is to create an autonomous
00:15:33
system that can handle different
00:15:34
situations across the globe and those
00:15:38
situations would help eventually assist
00:15:40
humans to lead a better life so the
00:15:42
basic iot CPS framework consists of
00:15:45
smart objects and the connections
00:15:47
between them then uh communication
00:15:50
networks are very important here come
00:15:52
continuous analysis of big data over
00:15:54
these platforms demand an efficient and
00:15:56
reliable Network structure
00:15:59
virtualization of nearly every physical
00:16:01
thing imposes a big Challenge on the
00:16:03
network service providers so there
00:16:06
should be Advanced Wireless technologies
00:16:08
that can handle such enormous eruption
00:16:10
of devices smart devices need a smart
00:16:13
Network
00:16:15
infrastructure artificial intelligence
00:16:17
and iot CPS how they are actually
00:16:20
working in the system is that we know
00:16:22
that the first Industrial Revolution
00:16:24
happened during 1760 to
00:16:26
1840 and during that revolution
00:16:29
industrial revolution a rapid growth of
00:16:32
machines took place in the Second
00:16:34
Industrial Revolution 1870 to 1914
00:16:37
people became richer in urban uh
00:16:41
currently what is happening is a smart
00:16:43
or a cyber evolution is underway a
00:16:45
number of interdisciplinary Technologies
00:16:47
and Sciences are converging and giving
00:16:49
rise to smarter softwares newer
00:16:52
materials uh very flexible and uh Robos
00:16:56
then groundbreaking inventions like 3D
00:16:59
printers and a whole range of
00:17:00
personalized web
00:17:02
services so AI enabled iot CPS would
00:17:06
basically be a good machine Learning
00:17:09
System uh in order to deal such big data
00:17:12
and what such a system would require is
00:17:15
data preparation capabilities learning
00:17:18
algorithms basic and advanced both
00:17:21
Automation and adaptive processes
00:17:24
scalability then emble modeling and
00:17:27
real-time decision making
00:17:29
in a cognitive Ai and iot CPA system
00:17:33
cognition as we earlier also discussed
00:17:35
is a process of acquiring knowledge and
00:17:37
understanding through thoughts then
00:17:39
experience and senses so intuitively
00:17:42
cognitive iot can be thought of as an
00:17:44
extension of iot which is capable of
00:17:47
understanding reasoning and
00:17:50
learning when we talk of air enabled iot
00:17:53
CPS Energy utilization routing or
00:17:56
traffic handling cost savings these are
00:17:58
some of the applications challenges in
00:18:01
this system would be as we see in the
00:18:03
recent iot trends that data is coming at
00:18:05
a faster speed from various sources in
00:18:07
various forms so basically it obviously
00:18:10
exceeds the abilities of an information
00:18:12
system to embi store and analyze and
00:18:15
process it the field of AI needs more
00:18:18
development with the Advent of such iot
00:18:21
CPS
00:18:22
infrastructure security is another major
00:18:24
concern as in this age of Technology
00:18:27
there will be cyber wall
00:18:29
we already see that cyber security is a
00:18:31
big Challenge and it is even impeding
00:18:34
National Security
00:18:35
sometimes so challenges of this CPS
00:18:39
include a number of specific challenges
00:18:41
which get raised due to the large scale
00:18:43
nature of this CPS system any physical
00:18:46
system may be predicted partially by
00:18:48
simulation but for a cyber process it is
00:18:51
difficult to predict the behavior this
00:18:53
difficulty to predict the behavior is
00:18:55
one big challenge when the Two Worlds
00:18:57
come together in a CPS they need to
00:18:59
operate in synchron they have to be
00:19:02
synchronized it poses a big challenge
00:19:04
for the global system control there are
00:19:07
even bigger challenges with the big data
00:19:10
so when we talk of challenges of iot the
00:19:12
Bas basic challenges are connectivity
00:19:15
security and Trust
00:19:18
interoperability scale of the whole
00:19:20
system then energy and environment so
00:19:23
these challenges are impeding the growth
00:19:25
of iot systems when we talk of data
00:19:28
analysis which is so so important in
00:19:30
these systems the challenges which are
00:19:32
impeding data analysis is analyzing the
00:19:35
massive and fast flowing data which
00:19:37
requires Advanced Technologies like
00:19:39
virtualization softwares then adaptable
00:19:42
cloud computing Etc it also needs very
00:19:45
powerful high performance Computing
00:19:47
devices that can provide the mechanism
00:19:50
to discover the underlying insights in
00:19:52
Big
00:19:53
Data so to conclude our talk we can hope
00:19:57
that in future future people will be
00:19:59
wearing intelligent gadgets eating
00:20:01
intelligent capsules that judge the
00:20:03
impact of the medicine on the body
00:20:05
living inside intelligent homes and so
00:20:08
on with the right policies we can get
00:20:10
the best of both worlds that is
00:20:12
automation without rampant Unemployment
00:20:15
uh with this we come to the end of the
00:20:17
session uh thank you all for a very
00:20:18
patient
00:20:21
[Music]
00:20:27
hearing
