系統設計者的話：學英文根本不用背，你講中文，你有在背中文嗎？學英文也是一樣的道理，那是自然而然就會的東西，用背的方式最後肯定都是不行的。 更多...

最好的學習就是有英文的父母，不斷的跟你說話、耐心的教導你，時間長了，你也就慢慢就融入其中了。

但是我們沒有英文父母怎麼辦？看影片學英文就學的起來嗎？學不起來。

我們從小到大，也看過了不少英文電影，至少也看過1、200部也跑不掉，但是為什麼看了這麼多的英文電影，英文還是很爛呢？

主要的原因是我們沒有從基礎性的強化學習。舉個例子，西方的幼兒出生時，他至少聽了各種大人講英文至少聽了2年，才慢慢開始說英文、與大人互動。

而且都是一字一句慢慢的說、慢慢的教、漸進式的、慢慢的、越來越習慣、越來越聽的懂、越來越會說的過程。

那麼我們看的影片、電影過程中，都是大人成熟型的講話速度、方式、語調，我們怎麼可能跟的上？

一旦跟不上，思想中就會產生覺的好難哦、我不行種種念頭，那麼產生了這種潛在式的排斥心理作用，那麼你就更難學的會了。


所以我們必須打好基礎、深厚的基礎，萬丈高樓平地起，唯有打下深厚的基礎，後面你走的路才會越來越順。

程式界裏有句話，程式債欠的越多，最後就只能放棄。學英文也是一樣，英文單字債欠的越多，你就直接上來文法、句型、語調，最後也只能放棄。

基於上述的思維，我打造了這套系統，從最基礎的單字學習起，不斷的強化、加深你對這個單字的聽力，透過各式各樣的影片、各式各樣的人、各種不同的場景。

就只是為了訓練一個單字的聽力，無數的人跟你講同一個單字，講個20遍、50篇、100遍(且不重覆)，我就不信學不起來。


這套系統做出來後，我自己在使用的過程中，確實真的能聽出來他們講的這個單字，而且他們講的速度再快也能聽的出來。

能聽的出來，就知道他講的大概意思；能夠知道他講的大概意思，就能夠用幾個簡單的單字回應；能夠回應，也就能夠對話；當能夠對話，就能夠再修正文法、句型、語調。


我的目地就是希望自己至少能聽的懂人家在說什麼，用英文跟人家對話，大家共同努力、學習。若有其它想法或對這套系統設計的建議，也可以與我聯絡：ichich2013@gmail.com

Machine Learning with TensorFlow (GDD Europe '17)
觀看次數：693
tensorflow
i'll
start
by
introducing
you
latency
predictions
so
let's
have
a
variations
like
breeds
poses
and
then
the
neuron
and
they're
all
connected
and
dog
images
they're
all
labeled
with
neurons
are
just
randomly
initialized
so
of
the
connections
between
the
neurons
weights
attached
in
the
neurons
about
basically
what
nura
neural
networks
used
weren't
that
many
layers
and
the
end
categories
I
think
it's
like
17000
the
early
networks
that's
what
we
mean
outperform
the
earlier
hand-built
hand-tuned
models
things
like
delaying
or
process
that's
called
back
propagation
cpu
and
tensor
flow
runs
on
cpus
and
devices
and
now
even
raspberry
pi
device
we've
been
serving
tensor
flow
graphs
to
build
a
tensor
flow
graph
and
it's
tensorflow
is
open-source
there
are
a
co-workers
on
the
tensorflow
serving
serving
is
a
very
very
high-performance
real-world
deployment
that's
one
of
the
production-ready
it's
things
like
this
this
is
one
of
the
visualizers
that's
one
is
showing
a
clustering
of
handwritten
digits
that
a
model
learned
visualizing
what's
happening
in
a
model
and
then
trying
to
debug
predictions
traditionally
been
a
very
difficult
tensorflow
serving
that's
something
that
different
tensorflow
the
open-source
community
really
sourcing
that
and
that's
the
end
of
it
thousand
external
contributors
on
tensorflow
how
best
to
use
tensorflow
high-level
api's
over
time
to
make
it
all
of
the
nuances
and
differences
in
word
lends
feature
to
the
Translate
app
outputs
a
human
sounding
caption
it
also
this
case
it's
not
just
classifying
the
there
and
there's
an
in-depth
post
about
condition
called
diabetic
retinopathy
an
ophthalmologist
they
take
an
image
signs
of
diabetic
retinopathy
that's
there
aren't
enough
ophthalmologists
to
ophthalmologist
at
diagnosing
the
architectures
are
good
for
solving
tensorflow
meant
for
like
big-budget
blockbuster
cucumbers
order
using
tensorflow
along
cucumbers
in
nine
different
categories
and
an
array
of
webcams
to
the
raspberry
the
belt
its
imaged
by
the
webcams
its
classified
and
it's
sorted
automatically
distributes
them
among
the
processors
the
estimator
api
which
is
a
high-level
what
we
call
the
canned
estimators
with
a
canned
estimator
you
can
create
a
deep
neural
network
classifier
in
literally
the
high
level
estimator
api's
you
get
a
we've
started
publishing
benchmarks
of
configurations
here's
the
cloud
tpu
that
it's
the
second-generation
tensor
teraflops
individually
but
they're
meant
time
into
these
tpu
pods
one
pod
is
115
one-eighth
of
one
of
these
tpu
pots
and
we've
been
working
on
a
compiler
that
converts
those
graphs
that
I
talked
on
tensorflow
lite
which
is
a
whole
new
android
mobile
devices
the
idea
is
that
general-purpose
code
when
you
combine
that
we're
calling
federated
learning
daunting
very
daunting
and
one
of
the
tensorflow
org
because
there's
a
getting
hands-on
introduction
to
tensorflow
as
like
a
convolutional
Network
which
is
architectures
and
parameters
at
playground
tensorflow
org
you
can
try
varying
the
number
of
layers
or
the
how
neural
networks
work
by
seeing
the
using
that
high-level
estimator
and
canned
estimator
api's
because
you
get
saved
and
restored
checkpoints
exporting
estimators
and
the
canned
estimators
are
automatically
and
at
geo
/tp
you
sign
up
there
is
a
sign-up
form
if
you're
the
tensor
flow
research
cloud
so
we're
really
state-of-the-art
research
so

相關影片

加Line好友 每日8AM 推播1單字

加入好友

其他人正在學習的單字

1. all
2024-05-02 21:21
2. ideal
2024-05-02 21:21
3. recommended
2024-05-02 21:21
4. healthy
2024-05-02 21:20
5. action
2024-05-02 21:20
6. guess
2024-05-02 21:20
7. food
2024-05-02 21:20
8. how
2024-05-02 21:20
9. first
2024-05-02 21:20
10. video
2024-05-02 21:20
11. dandy
2024-05-02 21:20
12. green
2024-05-02 21:20
13. students
2024-05-02 21:19
14. thinks
2024-05-02 21:19
15. hinges
2024-05-02 21:19

本月觀看影片單字排行榜

本月查閱單字排行榜

本月聆聽單字排行榜