Trajectory

s0

s1

s2

a0

a1

...

Before executing a

What do you know?

Prob(sj | si, a),

Prob(sk | si, a),

Prob(sl | si, a), ...

MDP Model of Agency

si

sj

sk

sl

a

Agent consults policy to determine what to doObjective: find policy that maximizes value function over finite horizon (or discounted ?)

Previous slide

Next slide

Back to first slide

View graphic version