Formulae Sheet: Markov Decision Processes

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 1

FormulaeSheet

MarkovDecisionProcesses

BellmanEquations: U (s) = R(s, (a)) + T (s, (a), s)U (s)


s

U (s) = maxaR(s, a) + T (s, a, s)U


(s)
s

NeuralNetworks
PerceptronLearningRule:
new_wi=old_wi+wi
where
wi=(tkok)xik

SigmoidActivationFunction: S(t) = 1+e1 t

DecisionTrees

GiniIndexforgivennode(t): GINI(t) = 1 [p(j|t)]2


k

SplittingbasedonGINI: GINI split = ni GINI(i)


i=1

Entropyatagivennode(t): Entropy(t) = p(j|t)logp(j|t)


j

InformationGain: GAIN split = Entropy(p) ( ni Entropy(i))


i=1

GainRatio:GainRAT IOsplit =

GAIN split
SplitINFO

i=1

Misclassificationerror: Error(t) = 1 maxiP (i|t)

Convertingtologbase2:
Ifyourcalculatordoesnothavethelog2function,hereshowtocalculate:

log2a = loga
log2

Note:logaislog10aandlog2islog102

OR

SplitINF O = ni log ni

log2a =

lna
ln2

You might also like