<s>
DeepDream	B-Application
is	O
a	O
computer	B-Application
vision	I-Application
program	O
created	O
by	O
Google	B-Application
engineer	O
Alexander	O
Mordvintsev	O
that	O
uses	O
a	O
convolutional	B-Architecture
neural	I-Architecture
network	I-Architecture
to	O
find	O
and	O
enhance	O
patterns	O
in	O
images	O
via	O
algorithmic	O
pareidolia	O
,	O
thus	O
creating	O
a	O
dream-like	O
appearance	O
reminiscent	O
of	O
a	O
psychedelic	O
experience	O
in	O
the	O
deliberately	O
overprocessed	O
images	O
.	O
</s>
<s>
Google	B-Application
's	I-Application
program	O
popularized	O
the	O
term	O
(	O
deep	O
)	O
"	O
dreaming	O
"	O
to	O
refer	O
to	O
the	O
generation	O
of	O
images	O
that	O
produce	O
desired	O
activations	O
in	O
a	O
trained	O
deep	O
network	O
,	O
and	O
the	O
term	O
now	O
refers	O
to	O
a	O
collection	O
of	O
related	O
approaches	O
.	O
</s>
<s>
The	O
DeepDream	B-Application
software	O
,	O
originated	O
in	O
a	O
deep	O
convolutional	B-Architecture
network	I-Architecture
codenamed	O
"	O
Inception	O
"	O
after	O
the	O
film	O
of	O
the	O
same	O
name	O
,	O
was	O
developed	O
for	O
the	O
ImageNet	O
Large-Scale	O
Visual	O
Recognition	O
Challenge	O
(	O
ILSVRC	O
)	O
in	O
2014	O
and	O
released	O
in	O
July	O
2015	O
.	O
</s>
<s>
The	O
dreaming	O
idea	O
and	O
name	O
became	O
popular	O
on	O
the	O
internet	O
in	O
2015	O
thanks	O
to	O
Google	B-Application
's	I-Application
DeepDream	B-Application
program	O
.	O
</s>
<s>
Related	O
visualization	O
ideas	O
were	O
developed	O
(	O
prior	O
to	O
Google	B-Application
's	I-Application
work	O
)	O
by	O
several	O
research	O
groups	O
.	O
</s>
<s>
After	O
Google	B-Application
published	O
their	O
techniques	O
and	O
made	O
their	O
code	O
open-source	B-License
,	O
a	O
number	O
of	O
tools	O
in	O
the	O
form	O
of	O
web	O
services	O
,	O
mobile	O
applications	O
,	O
and	O
desktop	O
software	O
appeared	O
on	O
the	O
market	O
to	O
enable	O
users	O
to	O
transform	O
their	O
own	O
photos	O
.	O
</s>
<s>
The	O
software	O
is	O
designed	O
to	O
detect	B-General_Concept
faces	I-General_Concept
and	O
other	O
patterns	O
in	O
images	O
,	O
with	O
the	O
aim	O
of	O
automatically	O
classifying	O
images	O
.	O
</s>
<s>
This	O
can	O
be	O
used	O
for	O
visualizations	O
to	O
understand	O
the	O
emergent	O
structure	O
of	O
the	O
neural	O
network	O
better	O
,	O
and	O
is	O
the	O
basis	O
for	O
the	O
DeepDream	B-Application
concept	O
.	O
</s>
<s>
This	O
reversal	O
procedure	O
is	O
never	O
perfectly	O
clear	O
and	O
unambiguous	O
because	O
it	O
utilizes	O
a	O
one-to-many	B-Application
mapping	O
process	O
.	O
</s>
<s>
The	O
optimization	O
resembles	O
backpropagation	B-Algorithm
,	O
however	O
instead	O
of	O
adjusting	O
the	O
network	O
weights	O
,	O
the	O
weights	O
are	O
held	O
fixed	O
and	O
the	O
input	O
is	O
adjusted	O
.	O
</s>
<s>
The	O
cited	O
resemblance	O
of	O
the	O
imagery	O
to	O
LSD	B-Application
-	O
and	O
psilocybin-induced	O
hallucinations	B-General_Concept
is	O
suggestive	O
of	O
a	O
functional	O
resemblance	O
between	O
artificial	O
neural	O
networks	O
and	O
particular	O
layers	O
of	O
the	O
visual	O
cortex	O
.	O
</s>
<s>
Neural	O
networks	O
such	O
as	O
DeepDream	B-Application
have	O
biological	O
analogies	O
providing	O
insight	O
into	O
brain	O
processing	O
and	O
the	O
formation	O
of	O
consciousness	O
.	O
</s>
<s>
The	O
DeepDream	B-Application
model	O
has	O
also	O
been	O
demonstrated	O
to	O
have	O
application	O
in	O
the	O
field	O
of	O
art	O
history	O
.	O
</s>
<s>
DeepDream	B-Application
was	O
used	O
for	O
Foster	O
the	O
People	O
's	O
music	O
video	O
for	O
the	O
song	O
"	O
Doing	O
It	O
for	O
the	O
Money	O
"	O
.	O
</s>
<s>
In	O
2017	O
,	O
a	O
research	O
group	O
out	O
of	O
the	O
University	O
of	O
Sussex	O
created	O
a	O
Hallucination	B-General_Concept
Machine	O
,	O
applying	O
the	O
DeepDream	B-Application
algorithm	O
to	O
a	O
pre-recorded	O
panoramic	O
video	O
,	O
allowing	O
users	O
to	O
explore	O
virtual	O
reality	O
environments	O
to	O
mimic	O
the	O
experience	O
of	O
psychoactive	O
substances	O
and/or	O
psychopathological	O
conditions	O
.	O
</s>
<s>
They	O
were	O
able	O
to	O
demonstrate	O
that	O
the	O
subjective	O
experiences	O
induced	O
by	O
the	O
Hallucination	B-General_Concept
Machine	O
differed	O
significantly	O
from	O
control	O
(	O
non-	O
‘	O
hallucinogenic’	O
)	O
videos	O
,	O
while	O
bearing	O
phenomenological	O
similarities	O
to	O
the	O
psychedelic	O
state	O
(	O
following	O
administration	O
of	O
psilocybin	O
)	O
.	O
</s>
<s>
In	O
2021	O
,	O
a	O
study	O
published	O
in	O
the	O
journal	O
Entropy	O
demonstrated	O
the	O
similarity	O
between	O
DeepDream	B-Application
and	O
actual	O
psychedelic	O
experience	O
with	O
neuroscientific	O
evidence	O
.	O
</s>
<s>
The	O
authors	O
recorded	O
Electroencephalography	B-Application
(	O
EEG	B-Application
)	O
of	O
human	O
participants	O
during	O
passive	O
vision	O
of	O
a	O
movie	O
clip	O
and	O
its	O
DeepDream-generated	O
counterpart	O
.	O
</s>
<s>
They	O
found	O
that	O
DeepDream	B-Application
video	O
triggered	O
a	O
higher	O
entropy	O
in	O
the	O
EEG	B-Application
signal	O
and	O
a	O
higher	O
level	O
of	O
functional	O
connectivity	O
between	O
brain	O
areas	O
,	O
both	O
well-known	O
biomarkers	O
of	O
actual	O
psychedelic	O
experience	O
.	O
</s>
<s>
In	O
2022	O
,	O
a	O
research	O
group	O
coordinated	O
by	O
the	O
University	O
of	O
Trento	O
"measure[d]	O
participants’	O
cognitive	O
flexibility	O
and	O
creativity	O
after	O
the	O
exposure	O
to	O
virtual	O
reality	O
panoramic	O
videos	O
and	O
their	O
hallucinatory-like	O
counterparts	O
generated	O
by	O
the	O
DeepDream	B-Application
algorithm	O
...	O
following	O
the	O
simulated	O
psychedelic	O
exposure	O
,	O
individuals	O
exhibited	O
...	O
an	O
attenuated	O
contribution	O
of	O
the	O
automatic	O
process	O
and	O
chaotic	O
dynamics	O
underlying	O
their	O
decision	O
processes	O
,	O
presumably	O
due	O
to	O
a	O
reorganization	O
in	O
the	O
cognitive	O
dynamics	O
that	O
facilitates	O
the	O
exploration	O
of	O
uncommon	O
decision	O
strategies	O
and	O
inhibits	O
automated	O
choices.	O
"	O
</s>
