<s>
The	O
paradigm	O
has	O
been	O
inspired	O
by	O
the	O
well-established	O
concepts	O
of	O
transfer	B-General_Concept
learning	I-General_Concept
and	O
multi-task	B-General_Concept
learning	I-General_Concept
in	O
predictive	B-General_Concept
analytics	I-General_Concept
.	O
</s>
<s>
Multi-task	O
Bayesian	O
optimization	O
is	O
a	O
modern	O
model-based	O
approach	O
that	O
leverages	O
the	O
concept	O
of	O
knowledge	O
transfer	O
to	O
speed	O
up	O
the	O
automatic	O
hyperparameter	B-General_Concept
optimization	I-General_Concept
process	O
of	O
machine	O
learning	O
algorithms	O
.	O
</s>
<s>
The	O
method	O
builds	O
a	O
multi-task	O
Gaussian	B-General_Concept
process	I-General_Concept
model	O
on	O
the	O
data	O
originating	O
from	O
different	O
searches	O
progressing	O
in	O
tandem	O
.	O
</s>
<s>
Evolutionary	O
multi-tasking	O
has	O
been	O
explored	O
as	O
a	O
means	O
of	O
exploiting	O
the	O
implicit	B-Operating_System
parallelism	I-Operating_System
of	O
population-based	O
search	O
algorithms	O
to	O
simultaneously	O
progress	O
multiple	O
distinct	O
optimization	O
tasks	O
.	O
</s>
<s>
This	O
view	O
provide	O
insight	O
about	O
how	O
to	O
build	O
efficient	O
algorithms	O
based	O
on	O
gradient	B-Algorithm
descent	I-Algorithm
optimization	I-Algorithm
(	O
GD	O
)	O
,	O
which	O
is	O
particularly	O
important	O
for	O
training	O
deep	O
neural	O
networks	O
.	O
</s>
<s>
In	O
addition	O
,	O
the	O
concept	O
of	O
multi-tasking	O
has	O
led	O
to	O
advances	O
in	O
automatic	O
hyperparameter	B-General_Concept
optimization	I-General_Concept
of	O
machine	O
learning	O
models	O
and	O
ensemble	B-Algorithm
learning	I-Algorithm
.	O
</s>
