<s>
This	O
page	O
is	O
a	O
timeline	B-Application
of	I-Application
machine	I-Application
learning	I-Application
.	O
</s>
<s>
1980s	O
Rediscovery	O
of	O
backpropagation	B-Algorithm
causes	O
a	O
resurgence	O
in	O
machine	O
learning	O
research	O
.	O
</s>
<s>
Scientists	O
begin	O
creating	O
programs	O
for	O
computers	O
to	O
analyze	O
large	O
amounts	O
of	O
data	O
and	O
draw	O
conclusions	O
or	O
"	O
learn	O
"	O
from	O
the	O
results	O
..	O
Support-vector	B-Algorithm
machines	I-Algorithm
(	O
SVMs	B-Algorithm
)	O
and	O
recurrent	B-Algorithm
neural	I-Algorithm
networks	I-Algorithm
(	O
RNNs	O
)	O
become	O
popular	O
.	O
</s>
<s>
The	O
fields	O
of	O
computational	O
complexity	O
via	O
neural	B-Architecture
networks	I-Architecture
and	O
super-Turing	O
computation	O
started	O
.	O
</s>
<s>
2000s	O
Support-Vector	O
Clustering	O
and	O
other	O
kernel	B-Algorithm
methods	I-Algorithm
and	O
unsupervised	O
machine	O
learning	O
methods	O
become	O
widespread	O
.	O
</s>
<s>
2010s	O
Deep	B-Algorithm
learning	I-Algorithm
becomes	O
feasible	O
,	O
which	O
leads	O
to	O
machine	O
learning	O
becoming	O
integral	O
to	O
many	O
widely	O
used	O
software	O
services	O
and	O
applications	O
.	O
</s>
<s>
1805	O
Discovery	O
Least	O
Square	O
Adrien-Marie	O
Legendre	O
describes	O
the	O
"	O
méthode	O
des	O
moindres	O
carrés	O
"	O
,	O
known	O
in	O
English	O
as	O
the	O
least	B-Algorithm
squares	I-Algorithm
method	I-Algorithm
.	O
</s>
<s>
The	O
least	B-Algorithm
squares	I-Algorithm
method	I-Algorithm
is	O
used	O
widely	O
in	O
data	B-Algorithm
fitting	I-Algorithm
.	O
</s>
<s>
The	O
techniques	O
later	O
become	O
known	O
as	O
Markov	O
chains.1943DiscoveryArtificial	O
NeuronWarren	O
McCulloch	O
and	O
Walter	O
Pitts	O
develop	O
a	O
mathematical	O
model	O
that	O
imitates	O
the	O
functioning	O
of	O
a	O
biological	O
neuron	O
,	O
the	O
artificial	B-Algorithm
neuron	I-Algorithm
which	O
is	O
considered	O
to	O
be	O
the	O
first	O
neural	O
model	O
invented	O
.	O
</s>
<s>
Turing	O
's	O
specific	O
proposal	O
foreshadows	O
genetic	B-Algorithm
algorithms	I-Algorithm
.	O
</s>
<s>
1951	O
First	O
Neural	B-Architecture
Network	I-Architecture
Machine	O
Marvin	O
Minsky	O
and	O
Dean	O
Edmonds	O
build	O
the	O
first	O
neural	B-Architecture
network	I-Architecture
machine	O
,	O
able	O
to	O
learn	O
,	O
the	O
SNARC	B-Algorithm
.	O
</s>
<s>
1957	O
Discovery	O
Perceptron	B-Algorithm
Frank	O
Rosenblatt	O
invents	O
the	O
perceptron	B-Algorithm
while	O
working	O
at	O
the	O
Cornell	O
Aeronautical	O
Laboratory	O
.	O
</s>
<s>
The	O
invention	O
of	O
the	O
perceptron	B-Algorithm
generated	O
a	O
great	O
deal	O
of	O
excitement	O
and	O
was	O
widely	O
covered	O
in	O
the	O
media	O
.	O
</s>
<s>
1967	O
Nearest	O
Neighbor	O
The	O
nearest	B-Algorithm
neighbour	I-Algorithm
algorithm	I-Algorithm
was	O
created	O
,	O
which	O
is	O
the	O
start	O
of	O
basic	O
pattern	O
recognition	O
.	O
</s>
<s>
1969	O
Limitations	O
of	O
Neural	B-Architecture
Networks	I-Architecture
Marvin	O
Minsky	O
and	O
Seymour	O
Papert	O
publish	O
their	O
book	O
Perceptrons	B-Algorithm
,	O
describing	O
some	O
of	O
the	O
limitations	O
of	O
perceptrons	B-Algorithm
and	O
neural	B-Architecture
networks	I-Architecture
.	O
</s>
<s>
The	O
interpretation	O
that	O
the	O
book	O
shows	O
that	O
neural	B-Architecture
networks	I-Architecture
are	O
fundamentally	O
limited	O
is	O
seen	O
as	O
a	O
hindrance	O
for	O
research	O
into	O
neural	B-Architecture
networks	I-Architecture
.	O
</s>
<s>
1970	O
Automatic	B-Algorithm
Differentiation	I-Algorithm
(	O
Backpropagation	B-Algorithm
)	O
Seppo	O
Linnainmaa	O
publishes	O
the	O
general	O
method	O
for	O
automatic	B-Algorithm
differentiation	I-Algorithm
(	O
AD	O
)	O
of	O
discrete	O
connected	O
networks	O
of	O
nested	O
differentiable	O
functions	O
.	O
</s>
<s>
This	O
corresponds	O
to	O
the	O
modern	O
version	O
of	O
backpropagation	B-Algorithm
,	O
but	O
is	O
not	O
yet	O
named	O
as	O
such.Griewank	O
,	O
Andreas	O
and	O
Walther	O
,	O
A	O
.	O
</s>
<s>
Principles	O
and	O
Techniques	O
of	O
Algorithmic	B-Algorithm
Differentiation	I-Algorithm
,	O
Second	O
Edition	O
.	O
</s>
<s>
1979	O
Discovery	O
Neocognitron	B-Algorithm
Kunihiko	O
Fukushima	O
first	O
publishes	O
his	O
work	O
on	O
the	O
neocognitron	B-Algorithm
,	O
a	O
type	O
of	O
artificial	B-Architecture
neural	I-Architecture
network	I-Architecture
(	O
ANN	O
)	O
.	O
</s>
<s>
Neocognition	B-Algorithm
later	O
inspires	O
convolutional	B-Architecture
neural	I-Architecture
networks	I-Architecture
(	O
CNNs	B-Architecture
)	O
.	O
</s>
<s>
1982	O
Discovery	O
Recurrent	B-Algorithm
Neural	I-Algorithm
Network	I-Algorithm
John	O
Hopfield	O
popularizes	O
Hopfield	B-Algorithm
networks	I-Algorithm
,	O
a	O
type	O
of	O
recurrent	B-Algorithm
neural	I-Algorithm
network	I-Algorithm
that	O
can	O
serve	O
as	O
content-addressable	B-Data_Structure
memory	I-Data_Structure
systems	O
.	O
</s>
<s>
1986	O
Application	O
Backpropagation	B-Algorithm
Seppo	O
Linnainmaa	O
's	O
reverse	O
mode	O
of	O
automatic	B-Algorithm
differentiation	I-Algorithm
(	O
first	O
applied	O
to	O
neural	B-Architecture
networks	I-Architecture
by	O
Paul	O
Werbos	O
)	O
is	O
used	O
in	O
experiments	O
by	O
David	O
Rumelhart	O
,	O
Geoff	O
Hinton	O
and	O
Ronald	O
J	O
.	O
Williams	O
to	O
learn	O
internal	O
representations	O
.	O
</s>
<s>
1988	O
Universal	B-Algorithm
approximation	I-Algorithm
theorem	I-Algorithm
Kurt	O
Hornik	O
proves	O
that	O
standard	O
multilayer	O
feedforward	O
networks	O
are	O
capable	O
of	O
approximating	O
any	O
Borel	O
measurable	O
function	O
from	O
one	O
finite	O
dimensional	O
space	O
to	O
another	O
to	O
any	O
desired	O
degree	O
of	O
accuracy	O
,	O
provided	O
sufficiently	O
many	O
hidden	O
units	O
are	O
available	O
.	O
</s>
<s>
1989	O
Discovery	O
Reinforcement	O
Learning	O
Christopher	O
Watkins	O
develops	O
Q-learning	B-Algorithm
,	O
which	O
greatly	O
improves	O
the	O
practicality	O
and	O
feasibility	O
of	O
reinforcement	O
learning	O
.	O
</s>
<s>
1989	O
Commercialization	O
Commercialization	O
of	O
Machine	O
Learning	O
on	O
Personal	O
Computers	O
Axcelis	O
,	O
Inc	O
.	O
releases	O
Evolver	B-General_Concept
,	O
the	O
first	O
software	O
package	O
to	O
commercialize	O
the	O
use	O
of	O
genetic	B-Algorithm
algorithms	I-Algorithm
on	O
personal	O
computers	O
.	O
</s>
<s>
1992	O
Achievement	O
Machines	O
Playing	O
Backgammon	B-Application
Gerald	O
Tesauro	O
develops	O
TD-Gammon	B-Application
,	O
a	O
computer	O
backgammon	B-Application
program	O
that	O
uses	O
an	O
artificial	B-Architecture
neural	I-Architecture
network	I-Architecture
trained	O
using	O
temporal-difference	O
learning	O
(	O
hence	O
the	O
'	O
TD	O
 '	O
in	O
the	O
name	O
)	O
.	O
</s>
<s>
TD-Gammon	B-Application
is	O
able	O
to	O
rival	O
,	O
but	O
not	O
consistently	O
surpass	O
,	O
the	O
abilities	O
of	O
top	O
human	O
backgammon	B-Application
players	O
.	O
</s>
<s>
1995	O
Discovery	O
Random	B-Algorithm
Forest	I-Algorithm
Algorithm	O
Tin	O
Kam	O
Ho	O
publishes	O
a	O
paper	O
describing	O
random	B-Algorithm
decision	I-Algorithm
forests	I-Algorithm
.	O
</s>
<s>
1995	O
Discovery	O
Support-Vector	B-Algorithm
Machines	I-Algorithm
Corinna	O
Cortes	O
and	O
Vladimir	O
Vapnik	O
publish	O
their	O
work	O
on	O
support-vector	B-Algorithm
machines	I-Algorithm
.	O
</s>
<s>
1997	O
Achievement	O
IBM	B-General_Concept
Deep	I-General_Concept
Blue	I-General_Concept
Beats	O
Kasparov	O
IBM	O
's	O
Deep	B-General_Concept
Blue	I-General_Concept
beats	O
the	O
world	O
champion	O
at	O
chess	O
.	O
</s>
<s>
1997	O
Discovery	O
LSTM	B-Algorithm
Sepp	O
Hochreiter	O
and	O
Jürgen	O
Schmidhuber	O
invent	O
long	B-Algorithm
short-term	I-Algorithm
memory	I-Algorithm
(	O
LSTM	B-Algorithm
)	O
recurrent	B-Algorithm
neural	I-Algorithm
networks	I-Algorithm
,	O
greatly	O
improving	O
the	O
efficiency	O
and	O
practicality	O
of	O
recurrent	B-Algorithm
neural	I-Algorithm
networks	I-Algorithm
.	O
</s>
<s>
1998	O
MNIST	B-General_Concept
database	I-General_Concept
A	O
team	O
led	O
by	O
Yann	O
LeCun	O
releases	O
the	O
MNIST	B-General_Concept
database	I-General_Concept
,	O
a	O
dataset	O
comprising	O
a	O
mix	O
of	O
handwritten	O
digits	O
from	O
American	O
Census	O
Bureau	O
employees	O
and	O
American	O
high	O
school	O
students	O
.	O
</s>
<s>
The	O
MNIST	B-General_Concept
database	I-General_Concept
has	O
since	O
become	O
a	O
benchmark	O
for	O
evaluating	O
handwriting	B-Application
recognition	I-Application
.	O
</s>
<s>
2002	O
Torch	B-Algorithm
Machine	O
Learning	O
Library	O
Torch	B-Algorithm
,	O
a	O
software	O
library	O
for	O
machine	O
learning	O
,	O
is	O
first	O
released	O
.	O
</s>
<s>
ImageNet	B-General_Concept
is	O
a	O
large	O
visual	O
database	O
envisioned	O
by	O
Fei-Fei	O
Li	O
from	O
Stanford	O
University	O
,	O
who	O
realized	O
that	O
the	O
best	O
machine	O
learning	O
algorithms	O
would	O
n't	O
work	O
well	O
if	O
the	O
data	O
did	O
n't	O
reflect	O
the	O
real	O
world	O
.	O
</s>
<s>
For	O
many	O
,	O
ImageNet	B-General_Concept
was	O
the	O
catalyst	O
for	O
the	O
AI	O
boom	O
of	O
the	O
21st	O
century	O
.	O
</s>
<s>
2010	O
Kaggle	B-Application
Competition	O
Kaggle	B-Application
,	O
a	O
website	O
that	O
serves	O
as	O
a	O
platform	O
for	O
machine	O
learning	O
competitions	O
,	O
is	O
launched	O
.	O
</s>
<s>
2011	O
Achievement	O
Beating	O
Humans	O
in	O
Jeopardy	O
Using	O
a	O
combination	O
of	O
machine	O
learning	O
,	O
natural	B-Language
language	I-Language
processing	I-Language
and	O
information	O
retrieval	O
techniques	O
,	O
IBM	O
's	O
Watson	B-Application
beats	O
two	O
human	O
champions	O
in	O
a	O
Jeopardy	O
!	O
</s>
<s>
2012	O
Achievement	O
Recognizing	O
Cats	O
on	O
YouTube	B-General_Concept
The	B-Application
Google	I-Application
Brain	O
team	O
,	O
led	O
by	O
Andrew	O
Ng	O
and	O
Jeff	O
Dean	O
,	O
create	O
a	O
neural	B-Architecture
network	I-Architecture
that	O
learns	O
to	O
recognize	O
cats	O
by	O
watching	O
unlabeled	O
images	O
taken	O
from	O
frames	O
of	O
YouTube	B-General_Concept
videos	I-General_Concept
.	O
</s>
<s>
2014	O
Leap	O
in	O
Face	O
Recognition	O
Facebook	B-Application
researchers	O
publish	O
their	O
work	O
on	O
DeepFace	O
,	O
a	O
system	O
that	O
uses	O
neural	B-Architecture
networks	I-Architecture
that	O
identifies	O
faces	O
with	O
97.35	O
%	O
accuracy	O
.	O
</s>
<s>
2014	O
Sibyl	O
Researchers	O
from	O
Google	B-Application
detail	O
their	O
work	O
on	O
Sibyl	O
,	O
a	O
proprietary	O
platform	O
for	O
massively	O
parallel	O
machine	O
learning	O
used	O
internally	O
by	O
Google	B-Application
to	O
make	O
predictions	O
about	O
user	O
behavior	O
and	O
provide	O
recommendations	O
.	O
</s>
<s>
2016	O
Achievement	O
Beating	O
Humans	O
in	O
Go	O
Google	B-Application
's	I-Application
AlphaGo	B-Application
program	O
becomes	O
the	O
first	O
Computer	B-Application
Go	I-Application
program	O
to	O
beat	O
an	O
unhandicapped	O
professional	O
human	O
player	O
using	O
a	O
combination	O
of	O
machine	O
learning	O
and	O
tree	O
search	O
techniques	O
.	O
</s>
<s>
Later	O
improved	O
as	O
AlphaGo	B-Application
Zero	I-Application
and	O
then	O
in	O
2017	O
generalized	O
to	O
Chess	O
and	O
more	O
two-player	O
games	O
with	O
AlphaZero	B-Application
.	O
</s>
