This transcript was autogenerated. To make changes, submit a PR.
Hello everyone, I'm suzy. See you again. Today we
will dive deep into you, the intersection of AI and IoT.
Before we just start over today's session, I would
like to extend my gratitude to the Kong 44
who invite me to be here and meet all of you again.
Now let's begin. If you didn't watch
my previous videos or recordings, these is a little
bit about me. I'm currently working as a DevOps manager
and software operations architect at signify
formerly famous lighting. Over the past three
years, I have deeply delved into and
involved into these AI and machine learning,
etc. Which is why I'm here.
And meanwhile, I'm also an irate contributor to
open source. So if you are curious, scan the QR
code on the left side where you will gain my
latest articles, speeches, and also my
upcoming technical books. So now,
first we'll discuss the AIot.
So what does it mean Iot?
Here Iot is. Iot is just a new term.
AI in Iot is Iot.
Here I just choose three examples.
It's quite practical examples for AI in
Iot. The first one, automated vehicles.
You may realize some brand is
quite famous, right? For example like Tesla
Wimble is kind of for pioneers in this industry
and also smart city. For example like
lighting, our company is a lighting leader
and radio surveillance and monitoring traffic
and smart buildings for manufacturing.
They take advantage like deep learning and deep
neural networks to do the
kind of product quality checks in production line.
And here I pick five use
case in real life AIop industry.
And for each part, for example like Niop,
it's a subfield for AI. Because for AI,
these have more than seven major
subfields. But here I pick five
to related AIot. So let's
dive into the first one.
Alexa, google Assistant, Siri, these are
all kind of virtual assistant. But you know,
in Iot we have some like a speaker, we can have
interaction with alexa, right? To light alexa,
this speaker to play videos or play
music or to implement
some tasks, et cetera. So they
take advantage like the NLP and also
for example like voice recognition model as well,
right? To achieve human being
to interact with this kind of IoT
product. Second, expert systems,
take IBI Morsen for example. It has been
hospital and what they can do like medical
diagnosis is quite amazing and
also very practical. And number three,
machine region, as I mentioned before, like neural
network and has been used for this
purpose. But here is machine region, it can be used in
production line to do this kind of production
quality checks. Number four, here it is,
planning is also one of these sub is one subfield
of AI self delving car.
They take advantage like applying AI technology to
navigate the routes. And it can help the driver.
It's a self driving car. IoT can help the driver to find
the best route. Number five
is quite amazing part.
For example like Boston dynamics sport robot. Have you
heard this news before? And it's really astonishing because
this robot can navigate terrain and also
to conduct like rescue missions
in dangerous situation. It will save human beings
life dramatically. It's really practical
in Iot. But if we think about
it for the Iot part, we take advantage.
AI technology, it's captivating.
It's like AI technology. It's like a
coin with two sides. Now we just explore amazing
part of AI. But think about, it's not
like a traditional software, right? If our traditional
software, if it has been hacked by AI, it will
just cause your website downtime
or your financial losses. But if it coming
to the Iot, for example like this robots
or some machine or something
else, or increase your flies, your bedrooms
temperature more than 100
degree celsius, it will kill people, right?
So we must prioritize the
Iot security, right? So now let's go
ahead AIot security. So we
will focus on how we can enhance security
in Iot. In AIot.
AI technology is
not new term. IoT has been coined in
1966, almost seven years ago.
But you know, with the explosion of
AI technologies, the current AI
security, the scope of AI security
will extend. It will expand very fast.
But we can still take advantage for the old security
classification, right? For example, these. It is.
Look at the diagram and you will see there's
an overlap between Iot security and information
security. Is network security.
And network security belongs to cybersecurity.
And look at the cybersecurity. I think you are quite familiar with the cybersecurity
a lot, right? We got hacked all kind of
related to the information security. And cybersecurity belongs to
information security. These. It is Iot security has
been included in cybersecurity. What is
Iot? What's a Iot security?
For example, like a 5G networks, Internet for
vehicles and also industrial Iot.
When we just talk about industrial control system,
sorry, related to Iot,
we definitely need to refer to this term ICI.
Iot means industrial control system.
So what is it? Look at the right side. There is a
timeline. Is ICI is a separate incident.
It's a real case happened since 19
three. Take one, for example, like 2012.
There is a campaign called Guys pipeline cyber intrusion
campaign. Think about it. This pipeline is
very crucial infrastructure or even a
nuclear industry. If they had
been kind of hacked by AI,
it will potentially not kill
a couple of persons. These will dramatically
destroy a city or even an entire country,
right. I think nobody will let
this kind of aiot out of our control.
Human beings control. So we definitely need to
prioritize Iot security as
much as possible, and also as soon as possible.
And when we talk about AIot
security, we definitely need to learn the AI security,
right? Because for Iot part
there is product, it's a real object just
send in front of you or just beside you.
And here is the AI technology,
or you can regard as AI model. These have interaction with
each other, the product and the AI part. Between AI
model and also these Iot object, our product
devices, et cetera. And we must make
sure the connection cybersecurity, right?
And also we need make sure the data security, right?
Yeah. These is between these AI model and
Iot devices. But we must make sure
for the AI part, also the IoT part,
both the security, they can make sure themselves
secure, right? So for AI part, here is these
is some list about AI security.
Like first data security, definitely privacy
preservation, because it's related to the AI part and
information security. And also explainability,
transparency. Because we
are kind of AI, this technology or model or machine
informed of human being is kind of a black box.
We must make sure everything expandable
and also transparent. Otherwise it has been trained
by, for example,
what should I say the very, by the purpose or
malicious purpose. Wow. You couldn't imagine this
machine can go, can take which direction, right?
Iot may be just do some very evil stuff.
We do not want this kind of thing happen into
the reality. And Iot security has
been included in AI security as well. And also
human AI interaction. Security model security
with the AI explosion, I mean, AI technology explosion,
we will see more and more model
will be launched. So definitely model
security and the best finance security. We are
human being living in
a world with different cultures, civilization and
even different religions, right? So we
found the machine. It's just a box. Oh, it's just a machine.
It's some. We are human being to give them
data, to train them to get the
result. What we are looking forward,
what do we expect, right? So we definitely make sure
there's a bears and fairness.
These barriers and fairness can be solved. These cannot be
exploited by malicious purpose and lifecycle
security and regulatory compliance, hybrid AI
human security. For each of these,
we definitely need to make sure the
AI part is secure and also, we will make sure our
Iot devices, for example,
like firmware and hardware,
as we all know, our apps or software,
they have bots. We definitely need make sure our
IoT product has kind
of a regular update and fix kind of
for potential vulnerability box,
right? And for the AI part, no matter,
we are in Iot, sorry, we are in
IoT industry or any industry,
for AI side,
for everyone. We need prioritize this AI security.
But today we talk about AIot, right? So for
the IoT part, we need to make sure the
regular updates and between both of them
and these cybersecurity and also the data security,
et cetera, right? So now
here's a summary. We talk about alt definition
and these real life use case in reality,
and also the information security IoT security
in AI and also AI security definition, right?
I think we just talk about all kinds of AIoT
stuff. Some very astonishing use
case it has been applied in reality.
And see, even for example, like that robotics can
conduct a rescue.
Okay? And so another part, we need to
care about the AI security very much because it's
not like software, right? It's some
object kind of interact with a
human being directly. And also the infrastructure
like nuclear industry,
ICIs, this kind of thing. So do
we definitely need to prioritize the
AI security, especially in IoT?
I hope you all enjoyed this talk and
maybe get a little bit inspiration. I will be very
grateful and thank you. See you soon.