Slides in PDF - of Marcus Hutter
Slides in PDF - of Marcus Hutter
Slides in PDF - of Marcus Hutter
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
<strong>Marcus</strong> <strong>Hutter</strong> - 94 - Universal Induction & Intelligence<br />
Brute-Force Approximation <strong>of</strong> AIXI<br />
• Truncate expectimax tree depth to a small fixed lookahead h.<br />
Optimal action computable <strong>in</strong> time |Y ×X | h × time to evaluate ξ.<br />
• Consider mixture over Markov Decision Processes (MDP) only, i.e.<br />
ξ(x 1:m |y 1:m ) = ∑ ν∈M w ν<br />
∏ m<br />
t=1 ν(x t|x t−1 y t ). Note: ξ is not MDP<br />
• Choose uniform prior over w µ .<br />
Then ξ(x 1:m |y 1:m ) can be computed <strong>in</strong> l<strong>in</strong>ear time.<br />
• Consider (approximately) Markov problems<br />
with very small action and perception space.<br />
• Example application: 2×2 Matrix Games like Prisoner’S Dilemma,<br />
Stag Hunt, Chicken, Battle <strong>of</strong> Sexes, and Match<strong>in</strong>g Pennies. [PH’06]