►
From YouTube: ECG+HTM (HTM Challenge Demo Video)
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
Hello,
I'm
kentaro
iska
I'll,
introduce
my
HCM
challenge
project
in
this
project.
I
try
to
detect
anomalous
parts
of
the
ECG
electrocardiogram
data.
This
project
is
still
in
prototype.
Honestly
speaking,
I
could
not
get
enough
accuracy
of
anomaly
detection
now,
but
I
think
this
project
shows
a
possibility
to
use
HTM
framework
for
biosensor
or
medical
application.
A
Okay,
look
into
the
project.
This
is
my
project
structure.
There
are
about
three
parts:
first,
collecting
data
using
ECC
device
with
simple
script.
This
script
correct
data
using
cellular
communication.
The
data
saved
to
the
CSV
file.
Second
part
is
the
FFT
fast
fourier
transform
part.
This
part
extracts
a
characteristic
of
the
ECG
data.
A
Sad
is
the
most
important
part,
detecting
a
memory
using
a
new
peak
in
this
project.
I
use
my
own
made
ECG
device.
This
is
the
device
this
device
can
read
about
190
samples
per
second,
it
is
enough
small
interval
for
allies
data.
This
devices
cost
is
under
fifty
dollar.
It
is
very
cheap,
corrected
data.
A
Okay,
this
video
shows
the
normal
ECG
data
learning
written
epic,
almost
animala
value-add,
0.2
20.4,
but
unmasked
are
going
to
high
regularly
the
high
animal
school
is
correspond
to
the
QRS
parts
of
the
ECG
data
in
ECG
data.
The
most
peak
and
spike
parts
called
QRS
wave.
You
big
temp
to
detect
normal
qrs
wave
at
animal
spot
after
enough
Ronnie
mushroom
also
help
experiment,
easy,
the
unmarried
addiction
with
music
and
approach
to
YouTube
that
we
do
also
shows
new
big
temp
to
mock
normal
qrs
wave
as
a
normal
spot.
A
Therefore,
I
need
another
strategy
to
deal
with
QRS
wave
in
new
peak
mailing
list.
Many
people
helped
me
to
a
nice
ECG
data.
Some
people
advised
me
to
use
Julio
transform.
Korea
transform
can
be
used
to
extract
characteristic
of
the
data
sequence.
I
wrote
a
fifty
combative
or
ECG
data.
Fft
output
is
the
vector
of
frequency
values.
So
I
got
the
time
series
of
vector
like
this
graph
data.
One
has
frequency
values.
Theta
2
has
also
frequency
values
and
so
on.
A
This
is
visualization
of
fft
combated.
Ecg
data
left
graph
is
a
normal
ECG.
Data
light
contain
animal
spot,
but
car
axis
is
time.
Step
and
horizontal
axis
is
a
frequency
step.
Cars
short
body
of
each
frequency,
you
can
see
non-uniform
part
in
light
graph,
actuary
non-uniform
part,
shows
animals
in
ECG
data.
A
A
A
Next,
I
use
a
lot
of
scarring
order.
I
a
describing
called
a
dynamically
to
sense
aprox.
It
can
laugh,
but
output
ulmer
score
are
strange,
a
low
risk.
You
are
going
to
high
regularly
I,
don't
know
what
caused
the
problem.
I
had
another
experiment
that
feels
better
in
color.
It's
deleted
only
scientific,
but
I
want
to
try
this
one.
So
I
put
back
this
encoder
from
past
commit
vector.
Encoder
is
only
one
skaara
encoder
instance
get
binary
representation
of
each
vector
value
and
combine
their
binary.
Do
presentations
vector
encoder
works
good.
A
The
output
seems
to
be
different
from
a
lot
of
scarring
coda.
For
my
condition,
this
is
victor
encoded
quit
you
can
see.
Animals
Potts
has
high
anomaly
score,
but
it's
not
in
a
high.
It's
only
3.1
to
0.15
now
I
continue
to
find
the
best
way
to
encode
time
series
FFT
vector
into
the
music.
If
you
are
interested
by
my
project,
I'm
happy,
my
github
repository
include
normal
and
anomalous
ECG
data,
so
you
can
try
to
use
my
projects
uncalled
for
your
experiment.