each G for the past fourteen years the ieee present of the ieee jack it cool be signal processing metal
the some portal or fun the refer to as the kill be metal
it's what i you refers to as a major metal
it was found about the ieee signal process society has been generously funded since its creation
but the texas instruments company
jack you'll be love to be in an engineer use was an inspiration to many do decided to pursue electrical
engineering as a career
well he is no longer an are males
his influence contain
as of already mentioned because it's a major metal
the ieee jack he'll be signal processing metal is presented by of the ieee
with the ieee medals and awards ceremony the will be held issue you're in august and san francisco california
however it is been the practise
of the metals down in society
that would be us
to present a special commemorative plaque
from the society
to the kill be metal B recipient
this year the recipient of the two thousand eleven ieee jackie a skill we signal processing metal
is in great do over shape
professor over a is receiving this honour or for pioneering contributions and you theory and applications of wavelets
and a filter
in could not be with us today we she will receive the metal at the ieee tripoli medal ceremony in
august in san francisco
but i do recommend we go for a round of applause in it since yeah
ladies and gentlemen that concludes today's ceremony i want to thank you all for attending the in the at triple
you signal processing society
order order B against than a that triple follows the recipient of the
ieee james flanagan at speech processing metal and the research you know of the ieee tripoli jet kill be metal
the so sorry hold as awards ceremony and really and i cast
a look forward to seeing you again next year
thanks again
i john for something thing all these works
and can grow to your workplace
so this is closing that
the official opening
and sort money
and that the and let me thank our musicians mister hour and mister of ski
for making this a a more you and
okay thank you
we are on time
it's miracle
the we are moving on to the technical program of for conference
the first one talk
i will come on the code um uh are or for a speaker
and a from a nokia
and professor before cory one and from helsinki university of technology
we will present or for a speaker
and should six
Q
where is send them and uh i have a great place or to introduce the an our to speaker
but that it really a in or was present or and that of market research and
nokia research center break was to enable a new business opportunities for not
yeah race the responsible for
you you know we're white that first
struck in the
and work closely we or not give a a in to promote open in the base
and working phone research is in collaboration with we we equal global research universities and
is
i believe that means that expense
what of time not only in is of this but also at the airports and a
and it well it's a T in computer science from the universal of think in one
and he joint here in
do got from for as the results go in software applications well
is previous positions include working at a at and T you go uh and uh us a visiting research scientist
at a so things
where can going to the wells role
called the also for email
a or and cool or able or for more than the or as would a i of
a a personal computer science social science statistics and holes like that
some of these papers are are are very can the school of single produce or so
including machine learning and and uh
uh
um
okay okay
is been also used to professor at to use C berkeley and stanford it works
is other interest include a a a a a long board surfing and scroll boarding
and i think that's why you like look in california would
i that the crib leads to work in a and risky more much of a is and i enjoy every
ladies and gentlemen of and it could be very oh do you know ahead
okay
good morning
um those of you who are right from uh uh U S i know
it's very early in the morning high i came from the your can the data from on but i'm
uh like uh
was mentioned i've spent a lot of times on the plane side actually not sure which time zone i'm
and in
um
what i was thinking about
what's the topic
that i would like to touch to for sets are how would i say
wide scope audience
i chose data
and they are multiple reasons of choosing data
one was that i spend
more than twenty five years some my life for searching for the medical data set
that could prove that my method some better than the other
um
a i for charlie that search um
of course
uh was a a a a difficult one not only because of the maybe the problems with my methods
but the access to data
and um
having worked long in the field
and and and looking at multiple datasets i know that
where we use of lee and
in that type of a situation ease the fitting
are methods to that particular datasets that we have a
or generating synthetic datasets which we know that will have their own
problems when when you use them some my feel was
a lot of what you would call machine learning or
information T legal learning or bayesian learning
one of the reasons is i i was so on the bayesian learning as we will hear um
later on during the conference was that as i had so little data at the bayesian
let's some what's uh easier methods to apply than the frequent just methods
so i decided to talk about data
today
just because i think it's a very common topic for
most of the things that you see the con for
uh
but i'm taking a very different approach spend again a a lot of time of doing um
the
i writing papers on on on on the different that a mathematical properties of of learning
and also implementing that different algorithms
and today i'm and going back to my roots of hacking
so i'm celebrating
this year forty years of hacking
which is a long time
and and taking a little bit just an perspective as people see
now so let's start
uh the the title to make sense of as that of by twelve well might wonder what those green things
that they are that the going it's not the warm gain the famous weren't gain that just to be in
in in the old days and on the mobile phones
so if you look at those uh green thing
a new reveal something out of eight actually it nice a description of representation all a mobile device is that
would be in a moving
the cool
going round that
white
yeah area are use the congress centre
so those would be the sensors that would be driving around of roads in in the device sees uh
that you carry with you
so
here really
uh typically nowadays which why call mobile computers because some of can be just and this so everything is a
computer to me
now this is
synthetic data
as we say
but that's
zoom a little bit
so this is real data
and this is real data
as collected by did not get devices
in prague
a by looking the navigation request
oh the different devices
in real time
you know
so pretty posing them on this plane with the court in eight
and the timing
i think it's see there is no matter
and the name
but after a while you see just thing developing nicely to that
map of prop
so
if we go for are
and assume i'll
this is a similar picture
of europe up
as seen
by billy in some cory is
all the all be
max
navigation
not just by the way there are some interesting cultural differences this speaker start mining this
as the topic of day
um
the green areas in fronds and also in russia
so that the uh the the those areas the people are more interested in navigation as opposed to seven your
of what the point of interest meeting
finding restaurants or
uh those are much more calm
so there's a multitude of day that that i've
face in my
current to life
it's also an a i the be do prove each to be an and here where this they are
and the C or or search of mine
is
starting to be fulfilled
we have access to data
in a a
totally
new way is
to to to develop of computers to to develop and of a multiple
different aspects that i that touched
so first i wanna uh
this costs
would you
some of the source is of this type of data that that is available to you
i my papers is to challenge you
later on to think about
what do you need to do
what do you need to think about when you in real life want to address
datasets like this
and you wise thing and and uh
so of making sense of the state
so first of all i showed you now
data that was based on the G P S
location information
in a time code
actually actually mobile device
already already like some multiple the for and
sensor data at the same time
we don't that they have a X are meters
yeah actually for me camera and audio is also a a a a a sensor
so each of these sensors that you at that is
this device
that that now so prevalent
adds a new layer
and there um more or more sensors that are coming a in the different radios that we are adding to
these devices
you're a more or more
a a sensor data that coming from the same source is
but at E
to do is uh uh base layer
of of location and time
why do why consider a location and time to be different than the ad
because
from my perspective
location and time can dish
the usefulness of a lot of the data
so if you don't know your location
many of the application for example like asking where the close
beer
part
don't make very much sense
so what some location and time of very fundamental
but beyond that
you can add more and more layers
and you in up in these piles of data
which of course
if you collapse them
shows how he man amount of data we have available
already from this type of the don't
that about it
the
well at least i it a part of my language but only row room full of geeks so we most
of us are
so excited about that the different numbers
i i was always possible
i i one studied mathematics that they are these different terms for the same numbers and and of four in
particular could be just as we are particularly excited of using
uh just power some two
so we talk about set up
now now set up what is an interesting number because it's the time um
currently approximately this year
although much of the data E is in fact copies of each or
we are producing a one point to set up of digital information
but it it this number also as any a large number
tends to be uh
difficult to to grasp
so um
it before sort of try do uh look at the different sources of this
at let's try to sort of reflect the little bit what it is so uh
okay
it's
approximately close enough to be six that we am
ten to the power of twenty one so we have twenty one zeros
but that doesn't tell us too much
so uh typically E um
if you want to just to get an idea of a big number or you should reflect it with uh
some major that you know
so let's
say that one but
of this would be one meter
and a good question to you is that now okay if
if we have
set meters
oh but this that's how long
what that this that's speech if we start from here
is it here from to moon
here to to peter
here to i'll for sent that already
okay so those have very clever and fast with the uh and know something about astronomy
would figure out that
this
is is that meters
is actually uh
the same as the diameter of the milky way
approximately
which is about hundred thousand like your
a lot of it's a big number
now i actually prefer somewhat the
the that bit more mundane uh reference sees that uh you can find on on on the net that when
we were a looking at the numbers
so one is that
set up by just amount of information if all the people on uh at
would be to twenty four seven four hundred year
or
it would be like seventy five really and
sixteen gigabytes
i had
for a of the data
which actually fills at four times
the more line time
uh but my favourite he's that if you like T V shows
it would be watching the the you know the he
C uh series or actual the first is sort of series
all the T V series twenty four
uh a four hundred and twenty five million years
continuous
talk about the sort of a
getting bored a bit probably
someone that
okay
now um
the standard answer when a
what we look at this type of a large dataset is that hey
we should be using a approximations we should be using sampling we should be doing you know
not exact things we we should be somehow
you know many lady
uh the the data set and then is that is correct
it's actually very old idea um in this scare out have lights this is a very old tab
eight thousand years old it
it's the babylonian tablet
that in fact
shows
a a approximation
all the uh the square root
a a a a a a unique
uh square as sorry at the diagonal don't our uh
although a unit square
uh
then
that allows us
in fact
to do
uh
a square with calculation
a for construction and
and and complex
so also to approximate measures sum of course of very all thing
now unfortunately
uh approach in a measures also as we know lose information
set K C that is perfectly fine
certain cases
uh it cost is by sees that we know what will be very hard
but the amount of data that we are talking about today
by four
requires as
to go to these approximate method of course
oh good have
like was pressing the wrong but
so let's look at a little bit about the source C east of the data now
at this is a
different picture than the previous ones
because this is not showing the absolute capacity
it is still the relative capacity of the type of the data that that is available for you go
so
in the old days when i remember one computer networks started
the remember forty years of hacking
one time
there's a lot of F T P traffic going on
E became very popular in the early eighties
uh there was something like telnet i don't know how many
remember or anything like that
and
but if T P was by far did don't mean a uh D that that was available meeting file transfer
from one place do not
that's to nineteen ninety
ninety only nineteen nineties
we all know um
the one of the
still annoying fact to the computer scientist that the physicist
so the introduced the the H U T P protocol and and the way but it was not the computer
this
should
and
the web was born
as only as you can see
the
if T P part
start
diminishing diminishing proportionally remember
this is a proportional of the act up some will not a you know a amounts are going up all
the time
and to way easy grabbing a more and more she
newsgroups are pretty happy
that tell let this sort of disappearing
email keeps it sort of a constant
and in that
if you go further
to dine T five
where has already captured half of the traffic
and you see in the upper corner or something interesting
like data
appearing from individual
peer-to-peer communication of the computer
which didn't used to be a case
in the past
because we only had this
few mainframes frames go
and if we go even further
to two thousand
we did see that the video
and video information
starts to grab a larger and larger here
oh the digital traffic
now way be still strong
but we D L the the purple part
uh ease
ease
just
morning
there at the corner
peer-to-peer
and web dominating
but
it's scroll very fast
so if we go no further
of course to two thousand five you see that the video is getting bigger
and um i don't know what a a because of the various different type of legally shoes another is used
the the uh the sort of percent each of a peer to peer
anyway not growing anymore to saying way and way is going down
and if you a ripe to two thousand ten
one could argue now that one the most important and interesting data sources that we and yeah
is the video
and nice there is no sign of it
going down at the moment
it actually
if you believe just go
which of course
uh can be a little biased you
the video will be so dominating at in the next couple of years on the network traffic that it will
be the majority of
and when
is in fact going down
proportion
which is quite obvious because you think about
the bits
required from of the normal web back
so when
the quality
sort that type of data
that that he's he's
moving around in these networks
is obviously you video data
so a lot of the things
that i used to be interested in which where relate with
sort of pattern matching in text
or some stuff in the files of music
is actually replace now in in stress to do this type of the mining
or a a processing of video
that that was the web
now
a king about where i just one of major something which it because it touched me so much in a
related to well sort talk about video
i wanna touch something um that a three D really captured my heart in in march in long beach ten
this was there raw is
from mit media lab
where on capturing ninety thousand hours of video all he's child growing up
and mining that we deal
uh in such a way that for example he could show
in the speed up manner
or the development of the work
water
in he a you know in the the the language development of the time
again a unique experiment
but related to topic
even more interestingly he's company as blue then
is working
and and delivering a
they are sort of a uh and now it takes or visualization
all uh both of the tv V broadcast real at with the social networking track
basically linking
something that is
shown on the T V on the discussion
that you have one on the net
okay
the little but different domain as a reference of this is the large had collider for those of you who
have not seen that
right picture
um
i do remember
that basically when the large had drunk lighter in and the data read was planned
there was a lot of talk about the capacity capacities now first of all a happened collider
has a hundred and fifty million sensors
so that's a lot of sensors an and we all know of course that these sensors are also proved using
then the data
with that in yeah
so of very rapid speech
so uh
the actual
approximate uh
so the bound of data with you about in this a structure used one paid the bite per second
i do remember that the original specs when we started doing the data rate i i i of was there
in the in this huge european union
consort you which has its benefits and a normal of your opinion and
"'cause" so a a is
uh we were talking about four point five that are bytes per second so
the the the bottom some got up
okay
hundred fifty million sensors
cool so
this is the physics six experiments
this is the science
big science you know
what does it do with you know the regular well or whatever it is this that very special device
expensive device put some
that that us come back this is now a picture of all of the whole well
related to the picture use of already and you know prod your up
um
the the different uh core
on
on the on the navigation this based on twenty billion court
now
remember what i showed you earlier
this a button one point two billion a devices currently on a that that of course and the number of
a mobile device is is a four point
or more than four billion
if each of these one point to build and device has ten sensor
it's ten
more than ten
bill
sensors
these ten billion sensors
although they don't feet
the sis that
with the same speed than a large how drum or would be
are still is
super substantial amount of
data that is available for
and this is
i'm not talking about the future somewhere
i'm talking about
the actual
today
not saying that all that sensor information is now at where collect the in one place
but it really really really
ease shown the potential and the different
uh
past that we have in the fit
i mean the different types of sensors in this mobile about computers
and i mentioned the sensors that are relay a with the a lot of the user in the phase or
or or very different types of a
uh uh uh uh a sort of uh positioning and so on
but this an interesting you
uh source
that at
to this sensor
wall
and that's the cognitive radio i so that the of some papers in a common give radio in this conference
as use one
to just want to point out that from this
that about uh perspective for those also you by the way to calm radio used in the
then and make a a location of the radio spectrum
uh in such a way that the device itself can actually choose
which part of the spectrum meet using
uh a it's signal
a transmission
uh actually can be used for a out of things do and and a for the sensor at of
so
that put detailed
introduction of county the radio will already bring that
a again at new very interesting source of since information which is in the infrastructure itself
so the traditional picture of having the device is talking to a
power
sell power and we'd already know something about the sting all strings that they can year
uh a how to sell power can recognise of nice that the device
is gonna change
to a picture which is a much more mesh
what a device are aware of each other's
press sense
or partially aware of each other's presence
in different type a radius spec
now these fingerprinting information ads and not the layer
again
which is inherent to the billy and
mobile about computer infrastructure that we have
well
and that the source of data
available for us all in that one is
ease the social media
there's currently about how nine hundred million social media users in the well
of course
if you look at that
um that
in principle means that there's is but something like one one point five billion of it's just social networks every
day
each of these base it's
leaves a trace or a is a a operation
and of course
if you want do uh divide this we know what that a majority of this is coming from a single
source
place book
seven hundred million currently
but the important part here is that this about thirteen billion
or more pieces of content axe
by these users
and this is the richest just context we have a one uh for for my because this is a uh
as you know in face book
or sort of information of all different types it is
it is both image we is it's is a low eighties textual data
E Ds a different lean C is sort of informative in a very different
additional things that you can uh a of course
C in the social space is that we have about sixty billion three
expected in two thousand eleven
and this sixty billion to
and uh
is
still a growing number because we have a four hundred sixty thousand you tweeter are guns at a daily which
by the is not to growth rate because
there are also people
that a drop that or accounts
but still shows that
the actual so the that population
is growing
and of course
back to our favourite
video
that is that a lot of the traffic
and in in the picture that showed you earlier
comes from you two
but uh of course in areas like in you dies states another it comes also from net flicks and so
so thirteen million hours of video on
in you to that doesn't look at very big number think about i was talking about the billions of there
but these to remember that these videos
are in fine it
snippets or so
so that two these thirteen billion our
is much more
uh in the number of videos that we have a below
um G
do
pushed on on but
so that million hours of video a it to you
thirty five hours a new video uploaded per
so those are you working on video mining
you have
great future
and no head of
now i'm i'm an optimistic and possibly person so i like this thing when all my life that things go
up and the are upper right corner
i like things growing i like things becoming more challenging i like things become fast there
small
bigger and so
what's the problem
well the problem is that
as opposed to you know
having this thing on the paper
what as to form a
or even ask calculation in your machine
we're talking about real systems you
and this data that exist somewhere
we need to access it we need do you hand the like and if you want to make use of
a
we need to be build systems that that that sort of a a a a a able to
do or what but
what happens if you are not careful of building this just a
that's a crack
this
actually uh the the the cover is from an older days but it to tall some ten we know that
that the one some point
decline in new york stock exchange in in a very short period of time
which actually result of a complex
a cohort it
yeah a computer software
that where of course doing uh uh uh what they are supposed to do they are competing on the market
in the super whom human human speech
the available data that they have
making it looking at weak signals and in sort of a a
uh a sort of a a or a the crash
so like i always point out
it's nice to write a paper
then have a good learning out
and a good predictive model
a bit to be
much more certain and when you start applying that
in the real well and you course
some interventions in the real world
the two very different
so what i want to talk about i was talking about the what now
what is the data available
i would like the little bit touch
how and why wide what want to do it for this audience i mean i'm not talking in operating system
conference i'm not talking in a networking conference on not talking about
uh the people even in my formal of those databases database
community
when i was a can be just signed this undergrad
i've as support about the memory computing trade
and a little bit later i was told about
the paging and you know virtual memory and and somewhat asked
my point is that
we are now
unfortunately in a different architecture
we are using a difference just a and when we are writing our algorithm
and when we are running them
we need to take into account
to a a to greedy D tells which are beyond the turing machine yeah even beyond the for neumann
sort of traditional computer model
we need to look at that aspect if we really wanna work in the real
with this data
uh this a be yeah
between
what is the practise shouldn't traits correct
in in in in in menu plating in of so called internet companies for example for the data
and the work we do and the very at once work we do
in a sophisticated methods of understanding that
sometimes these things
the gap is smaller
sometimes it's much lot
what a talk about
a a and you with
how a all the important aspects of this just then
that we should take into account
what we are writing our algorithm
when we are
you know building them for that the by well
not building them for the
uh
well like my favourite was the or dataset that a over fit it so badly
and list the city
one of the things that has changed
dramatic is that
when we you building things we don't need
bill them
a in such a way that we have to uh make sure that the maximum requirement is somehow con
because in the old days when you had a computer it had a certain amount of uh computing power
said amount of memory
and was a box some
no L this C is an should
of course that has sort of sneak in with clout
so L L this T allows you to use dynamic to computation power or
of a larger or a in today
is such a way that you don't have to
uh uh so the the
three D term mine you model the competing power do you use
no this is a to to the same
development of you had in the uh i guess a into that read data structure of course
where one has
six table set
and then the dynamic to
and you defined find a dynamic table and you don't have to care about the the that that you will
ever go wild
so this
feature where you need more data
and you just grad
more of the competing power
and then
at the same time of course this is meaningful for only because you have multiple users at the same time
sharing
this particular pull
you will goal uh to a much lower T
so in some sense this ls this city
has um
allowed us to do uh things which were not add all
uh feasible in in the past
it was not that long ago i think about
seven eight years ago
we were running multinomial pca out
a on on eight i if fixed cluster for a search and stuff actually doing sort of
uh search engine uh a probabilistic modeling all the the work of course is in in in a hundred million
documents or more
and a typical run
there was
deep limited by competing power that we have
so we we had to run like three weeks in a role
in a small cluster together get a model of like to fifteen million
or twenty male and documents
those you not be a a a no pca
or mode only piece you knows that it sort of a under you do something clever
it's actually quite computationally intensive thing
and and there was no way you doing any kind of dynamic things so all our experiments where
where sort of restrictive by to computing power at the university which was not that great
so we we had to work on
but the the L to sit B for as a historical remark it's to an to you very old knowledge
and
my
okay very uh quotes from the science fiction church or
uh as many of the computer size yeah ideas have a are very only in church or
is this
first
uh commercially sold
um
story by our to see clock
uh from this thing size fiction called
uh a a rescue party
and rescue party
most telling
about
at race
a call uh how do already a place called powered or with the race some power or
yeah had a collective mine
and depending on the problem in the universe
the that race
collected more minds
dynamically to solve the problem
so but in that today if you had to read a really big problem
telepathic connection
in the different places and the unit thus allowed to solve much harder problem
this is nine this was really really are
so fifty so the uh a sort of
or was a used how
well some of the site fix at uh and a or to reflects of a
of the future
a second aspect
well as a list this C you might think okay so it's the cloud stuff
is the robustness are
rocks this argument easy a very complex one uh because it's basically depend
on one um
what do you wanna do
so this is nice result by eric rule or from cow
and this that depending on
which time men she's you are interested in
meeting you're accessing some data
and you want to do a consistent access
or you should be able to access when partitions are allowed in the network
or yeah data should be always available if need
you concept set is five or all of these at the same time
of the
you can only go in in this
different
or warners or a different of borders of the trying is such a way that
if you wanna do for example search
uh you're actually very petition for and
at you have certain type of consistency but not all things are always a
or you bit or end where you don't care about consistency at all
you just
basically are doing
uh transfers but
you very tolerant for petitions as as those who want to legally
imposed against the this type of a part of C no
uh and things have very available
and on the other hand in this
just just do body data bases that many many years ago and is working on them the consistency and availability
a very important
but they were not very tolerant for partition
now why this is important this is important because we have talking about highly to but it just of remember
i was talking about one billion
devices
that does the sitting in different parts of the well
yeah can make it by different make words
a or actually allowing different types of but uh
uh mouth functions and there is them
there is no
consist then everything use up all the time
notion i all
so when you building you out rhythm
they cannot be based
bone
getting all the necessary information but
by necessity is you of course i all assume that you already figured out that in this is that the
by well
the outward have to be on line H
batch algorithms of taking
you know this twenty billion queries and running them and doing it
it usually not the way to go because the response times
uh for the problems that you want saul
i'm not
uh these
so we had a ct and robust
but the
the one thing that is so dear to my heart
these energy
so i
i preached just every place i'd bin
now for the last two years i'm preaching it here too
and i'm pointing out if five where a do that
i would go to this field of energy
efficient compute
so my argument is that the current architecture as we will see
these fun the mentally
imposing the similar even theoretical call uh sort of boundaries
and to to to computing ask we used to do with memory
and computing
energy
is this so
so let's look at the real life situation why of things become difficult it's it's stick exam O
or of a a a a processing that you
you know
normally you used to this type of about
processing where you have a single box now we are in a well what we a multiple devices that i
connected to each other
and you have your how or it your great algorithm and doing the video stuff
has a choice
where to go
where to execute
to let to look about this because it this problem
in particular because E a very import
if you to could the it look at the experience domains
it it's experience i ancient
you can of course running everything in in in the device itself
which is the you know the this type of that mobile computer you could do the video editing here
you get the video you at to here
uh you get and the display
well we know that uh in the case of a larger things
this will be very very very slow all the user experience will be pretty bad and then in addition there
might be some other
sort of a user experience issues that yeah but it's basically little
of course we can do peer to peer so we can steal somebody's else computing interesting idea
taking a little bit more computational power or from the neighbourhood
to do the video eating
fine
or
what we can of course to ease that we think all okay we have this
yeah last the city up there
we have the clout so we send the P deal
uh to uh the the actual data
up to the cloud to be processed there where you can the fast
process of course there are you add now talk
to transmit
so in the user experience die main should you one of the things that you want to use lee
from a user perspective
to uh
uh just optimized is the time
it and can't use it doesn't care where it happens you just
as soon that these meeting
we are living in a very be a
well
this is the pure experience per spec
if you look at the economics perspective
oh this one
again
if you to everything in the device itself not is that i now divided in that's a way that the
law were in use the device upper in these the cloud in in the middle this of transmission
somebody have to pay for the clout
so but in the the day it usually doesn't come from
from a from nowhere
so
you have to you have to pay for that economics of the cloud
but form a a a a again at a user perspective
you also paying on the transmission data com
in most places
though that will also be expressed
but then we come to the very fundamental question
if it do
the video at eating on the device
you're running out of energy in the car and
mode very fast and we all know already that that even taking pictures
not even at thing on the video well run out of our batteries very
so okay
not a very
not a very feasible thing
so what about if we just
put it on the clock
so i mean then we don't run out all sort
the next thing is do we can steal our neighbours energy
doesn't make you very popular
because the the person who want to make a call next time and doesn't have any energy anymore
probably doesn't like very much the idea that uh uh uh he or she has support your
video at T
if you put go to the cloud side
we all know the problems already that
this is like a
no free lunch situation
the cloud
server from use is also and it might be a again G
that is not immediately effect you
but as a total off
somebody will have a problem
with the in G and we're talking about a green data centres nowadays a lot
and
this also fundamental lame
which is an interesting question that how much can you concentrate
in in a one place because you see in there
bottom uh approach
you're introducing new energy source every time you're uh introducing a new device that does to computing
in the cloud
the low energy load grows linearly
at the and uh of the cloud side with respect to the customers it has to sit
but even more fundamentally
this trade
and for D in real life
tends to be controlled by the fact that sending bits
but much more energy than computing
part it is because of the the coding in that we have
you know we are so far from the
uh channel limit made that basically uh we are essentially
uh i have to put
much more power than we would need to
to send uh the but be to to correct the noise errors that we have
but basically
this balance is good to did date
almost all of the the the data manipulation i've been talking to do
the balance between what do you compute locally and you energy
where do you put where you have the energy of here
and then what's the transmission energy
did you are doing if you think about rendering for example watch screens directly from the class
and
i wanted to point out that um
academically
energy a a fundamental thing
in a G something you cannot chi
so it's very appealing theoretically
in real life systems
the the economics and experience
are the ones
that really also dictate what will be uh used in pratt
so my question
i sort of problems that i one lee would you
is that
basically
we know
and that shown that we have multiple sources of data available
the question is that
what's to a kid had talk so how do we capture parse and analyse on them on the fly meaning
that that one do things on wine
and radically different source sees this is um
uh you know the sense sorry fusion he's one term that people use
and different
commune his we use different terminology
to more me just the head of is attributes of source of that
the second question ease the architecture question
what what like take to build these socket to that actually a robust but
and have strong elastic properties
how do you right go out it means such a way that
and
even if you running in the signal much
i would write the running T with what one billion device
cross the uh
maybe with the cloud component
the third question
is that
how do we tackle this energy efficient computing
because in fact
and
like i say
and from the theoretical perspective almost like a the perspective that you had this after a four and model
you can look at energy as a
the different component there and do a lot of analysis
even a like we need to balance environment concerns but this is practicalities
and the in user experience
for this and just
now
why Y in or G so fundamental to me
it's fundamental for a reason that this first time
in in my lifetime
we are reaching the levels that then i can argue that i can foresee service sees
that had not be bill not because of the call reasons but because of the reason that we don't not
have enough energy on yeah
the runs at to
six billion user service
what a run on the device and what is run on on the on the back
so this is what i wanted to leave you a today
i just one to remind you that my
medic search
for the medical data sets
that quest actually has been filled
i have always think than things and and a well
exciting me there
lying there
and now i see that about twenty five years of my life
i haven't solved very many things related but this
this is
i've sold a lot of things would related but i don't think
would you guys in a community with the great P H student is that i've have the privilege to work
with
but basically
i guess said
i now know
that we are facing an air a
so there are almost need
to our exist then is that we as a research communities
need to address in a different way
okay
so
that's what i want to say to stop thank you very much for your attention
i Q for very exciting presentation
we we have a very
challenging research problems to work on with a whole community here
so problem
in in them
scientific uh uh uh four
as
for
global
a local uh
energy consumption and problem of the whole or
and uh we have a time for couples or questions so please
okay
as mike so
excellent
i
very much
oh
two
two we
a
so i know it's your
or good thing of be important in this environment
our privacy and security
and so i
i so the like
my own close
excess
there
have
yeah
for all network
we work
yes
and
i deliberately chose
not the say the work privacy and security
as somebody characterised rice me a long time ago that henry
if we reset aspect you always doing is violating
i
C really people's privacy and that's of them
and no i think it very seriously yeah i just decided to leave it for the question because i knew
that the question what car
and
of course
first of all
at so the that they can very sign this perspective
the more we get or information
the more i can reverse engineer
that's a fundamental that and i could even do it in the ways that
people think it uh you know you you and and mice thing sell you at run
noise but if you can predict the noise model you can reverse engineer a lot of stuff and you could
do very complex things
now
for me first of what privacy is always a trade off it strike a reliability really
this is certain pay all you get from something
and certain cost that you have
if the cost
is higher than the pay off you should not do it so the cost for your privacy
and and the aspect you should be able to first of all use and always be able to opt out
that that's that's a the the first
but the second i of point out that that learned that and never thought so much was that the or
a lot of things where you could do a a or trained
how to get and alice
without violating any kind of privacy ask
you basically
at hating like this traffic
sure sure the ear
that does that just totally and no anonymous
no idea who is their don't will the individual points are followed
in a sequence
yeah just point in the time in the data say you don't know that they coming from the same source
you can still do a meaningful alice
that's the first good
second good nice for research community that the are privacy preserving make is
that you can build
in this
yeah sort of put the traffic
ways of handling this thing so it's a future research brought
and
a third question is that
i usually do the channel see how many you guys in this room has actually to of your cookies
i mean just to be very popular
just to be very popular but also this some if fit of not turning of them more relays in this
and that again
so i E each very cultural location only what happens but i hope we you
and and my main point always is
uh and it is nice
you should know
and use should be able to opt out
but if this it twice
if you want to uh a a a a get some benefit out of that information that is available
security ease of different ish
i i think everybody shares the the the question of the security problems uh the concept of the security problems
that we have
and and D sub very complex is used again called zero even regular to issues
that we face a different ways in your a
a in us any nation
my lattes are working and i'm working on growth economies to
was than well and i can tell you that these issues a very different in different
mark
security uh protocols are a good research topic do i think the the privacy consent tends to be the higher
one among the people a we know all the problems that currently are
face books and google another side facing it is but don't
i decided not to it over in for size this because i was talking from a size perspective
and thus a recess perspective but we have to be of course it it just about that a a work
we do to
right right no short question
and hopefully a ask