ped an a

me year, DeepM

ort helpdesk, but also discuss morality, express opinions, and answer ge

mans as a t

ncingly int

at could n

l agent (A

trained a conve

omputer pr

wer have dr

r the past four

ate of acceleration

ike magic.

more and

a world fu

es feeling left behi

o does not un

vation this century. A

powerfully than

ape our future

tificial intelligence

ning matters

Why machine

he Appendix

ommendations i

look at

tc., take a

rojects to atte

extbooks to

which course

sted in figurin

ou're more in

hours.

concept

igh-level ma

g up-to-sp

s is a guide

ledge of th

iscussed,

ut it isn’t nec

sary to h

bra, and c

ramming, linea

bility, statisti

epts in

e. Basic con

ccessible to

is intended to b

who is curi

bout ho

h techni

rimer on m

echnic

ople who

to speed o

ho want to

echnical

curriculum.

ing your machine

curated

deep reinfor

ue learning p

esses. Q-lear

ing, policy lear

olution

m the brain. Co

ploration and e

ning works. D

Why, where, and

omposition (S

nalysis (PCA

rincipal compone

sionality re

ng: k-mean

nce and machine

earning — past,

tions, overfitt

earners:

on-paramet

SVMs.

ession an

f classifica

wo metho

ient des

erfitting, and gr

loss functions,

near regr

wer key. I

arning with an

he big pict

Roadmap

nations

Simple, plain-English

Humans

Learning for

Machine