On the Complexity of Best Arm Identification with Fixed Confidence
|
|
- Meghan Antonia Wilkinson
- 5 years ago
- Views:
Transcription
1 On the Complexity of Best Arm Identification with Fixed Confidence Discrete Optimization with Noise Aurélien Garivier, Emilie Kaufmann COLT, June 23 th 2016, New York Institut de Mathématiques de Toulouse LabeX CIMI Université Paul Sabatier, France Université Lille, CNRS UMR 9189 Laboratoire CRIStAL F Lille, France
2 The Problem
3 Best-Arm Identification with Fixed Confidence K options = probability distributions ν = (ν a ) 1 a K ν a F exponential family parameterized by its expectation µ a ν 1 ν 2 ν 3 ν 4 ν 5 At round t, you may: choose an option A t = φ t (A 1, X 1,..., A t 1, X t 1) ) {1,..., K} observe a new independent sample X t ν At so as to identify the best arm a = argmax a µ a as fast as possible: stopping time τ. and µ = max µ a a Fixed-budget setting Fixed-confidence setting given τ = T minimize E[τ] minimize P(â τ a ) under constraint P(â τ a ) δ 3
4 Best-Arm Identification with Fixed Confidence K options = probability distributions ν = (ν a ) 1 a K ν a F exponential family parameterized by its expectation µ a x 1 x 2 x 3 x 4 x 5 At round t, you may: choose an option A t = φ t (A 1, X 1,..., A t 1, X t 1) ) {1,..., K} observe a new independent sample X t ν At so as to identify the best arm a = argmax a µ a as fast as possible: stopping time τ. and µ = max µ a a Fixed-budget setting given τ = T Fixed-confidence setting minimize E[τ] minimize P(â τ a ) under constraint P(â τ a ) δ 3
5 Best-Arm Identification with Fixed Confidence K options = probability distributions ν = (ν a ) 1 a K ν a F exponential family parameterized by its expectation µ a x 1 x 2 x 3 x 4 x 5 At round t, you may: choose an option A t = φ t (A 1, X 1,..., A t 1, X t 1) ) {1,..., K} observe a new independent sample X t ν At so as to identify the best arm a = argmax a µ a as fast as possible: stopping time τ δ. and µ = max µ a a Fixed-budget setting Fixed-confidence setting given τ = T minimize E[τ δ ] minimize P(â τ a ) under constraint P(â τ a ) δ 3
6 Intuition Most simple setting: for all a {1,..., K}, ν a = N (µ a, 1) For example: µ = [2, 1.75, 1.75, 1.6, 1.5]. At time t: you have sampled n a times the option a your empirical average is X a,na arm 1 arm 2 arm 3 arm 4 arm 5 if you stop at time t, your probability of prefering arm a 2 to arm a = 1 is: P ( Xa,na > X 1,n1 ) = P ( Xa,na µ a ( X1,n1 µ 1 ) 1/n1 + 1/n a > ( ) µ 1 µ a = Φ 1/n1 + 1/n a ) µ 1 µ a 1/n1 + 1/n a e where Φ(u) u2 /2 = du u 2π 4
7 Uniform Sampling 5
8 Uniform Sampling 5
9 Uniform Sampling 5
10 Uniform Sampling 5
11 Uniform Sampling 5
12 Uniform Sampling 5
13 Uniform Sampling 5
14 Uniform Sampling 5
15 Uniform Sampling 5
16 Uniform Sampling 5
17 Uniform Sampling 5
18 Uniform Sampling 5
19 Uniform Sampling 5
20 Uniform Sampling 5
21 Uniform Sampling 5
22 Uniform Sampling 5
23 Uniform Sampling 5
24 Uniform Sampling 5
25 Uniform Sampling 5
26 Uniform Sampling 5
27 Intuition: Equalizing the Probabilities of Confusion Most simple setting: for all a {1,..., K}, ν a = N (µ a, 1) For example: µ = [2, 1.75, 1.75, 1.6, 1.5]. Active Learning You allocate a relative budget w a to option a, with w w K = 1. At time t: you have sampled n a w a t times the option a your empirical average is X a,na arm 1 arm 2 arm 3 arm 4 arm 5 if you stop at time t, your probability of prefering arm a 2 to arm a = 1 is: P ( Xa,na > X 1,n1 ) = P ( X a,na µ a ( X 1,n1 µ 1 ) 1/n1 + 1/n a > ( ) = Φ µ 1 µ a 1/n1 + 1/n a ) µ 1 µ a 1/n1 + 1/n a 6
28 Improving: trial 1 7
29 Improving: trial 1 7
30 Improving: trial 1 7
31 Improving: trial 1 7
32 Improving: trial 1 7
33 Improving: trial 1 7
34 Improving: trial 1 7
35 Improving: trial 1 7
36 Improving: trial 1 7
37 Improving: trial 1 7
38 Improving: trial 1 7
39 Improving: trial 1 7
40 Improving: trial 1 7
41 Improving: trial 1 7
42 Improving: trial 1 7
43 Improving: trial 2 8
44 Improving: trial 2 8
45 Improving: trial 2 8
46 Improving: trial 2 8
47 Improving: trial 2 8
48 Improving: trial 2 8
49 Improving: trial 2 8
50 Improving: trial 2 8
51 Improving: trial 2 8
52 Improving: trial 2 8
53 Improving: trial 2 8
54 Improving: trial 2 8
55 Improving: trial 2 8
56 Improving: trial 2 8
57 Improving: trial 2 8
58 Improving: trial 2 8
59 Improving: trial 2 8
60 Improving: trial 3 9
61 Improving: trial 3 9
62 Improving: trial 3 9
63 Improving: trial 3 9
64 Improving: trial 3 9
65 Improving: trial 3 9
66 Improving: trial 3 9
67 Improving: trial 3 9
68 Improving: trial 3 9
69 Improving: trial 3 9
70 Improving: trial 3 9
71 Improving: trial 3 9
72 Improving: trial 3 9
73 Improving: trial 3 9
74 Improving: trial 3 9
75 Improving: trial 3 9
76 Optimal Proportions 10
77 Optimal Proportions 10
78 Optimal Proportions 10
79 Optimal Proportions 10
80 Optimal Proportions 10
81 Optimal Proportions 10
82 Optimal Proportions 10
83 Optimal Proportions 10
84 Optimal Proportions 10
85 Optimal Proportions 10
86 Optimal Proportions 10
87 Optimal Proportions 10
88 Optimal Proportions 10
89 Optimal Proportions 10
90 How to Turn this Intuition into a Theorem? The arms are not Gaussian (no formula for probability of confusion) large deviations (Sanov, KL) You do not allocate a relative budget at first, but you use sequential sampling no fixed-size samples: sequential experiment tracking lemma How to compute the optimal proportions? lower bound, game The parameters of the distribution are unknown (sequential) estimation When should you stop? Chernoff s stopping rule 11
91 Exponential Families ν 1,..., ν K belong to a one-dimensional exponential family P λ,θ,b = { ν θ, θ Θ : ν θ has density f θ (x) = exp(θx b(θ)) w.r.t. λ } Example: Gaussian, Bernoulli, Poisson distributions... ν θ can be parametrized by its mean µ = ḃ(θ) : νµ := νḃ 1 (µ) Notation: Kullback-Leibler divergence For a given exponential family, d(µ, µ ) := KL(ν µ, ν µ ) = E X ν µ ] [log dνµ (X ) dν µ is the KL-divergence between the distributions of mean µ and µ. We identify ν = (ν µ1,..., ν µ K ) and µ = (µ 1,..., µ K ) and consider { } S = µ (ḃ(θ)) K : a {1,..., K} : µ a > max µ i i a 12
92 Back to: Equalizing the Probabilities of Confusion Most simple setting: for all a {1,..., K}, ν a = N (µ a, 1). For example: µ = [2, 1.75, 1.75, 1.6, 1.5]. You allocate a relative budget w a to option a, with w w K = 1. At time t, you have sampled n a w a t times option a and the empirical average is X a,na. if you stop at time t, your probability of prefering arm a 2 to arm a = 1 is: P ( Xa,na > X 1,n1 ) = P ( X a,na µ a ( X 1,n1 µ 1 ) 1/n1 + 1/n a > ( ) = Φ µ 1 µ a 1/n1 + 1/n a ) µ 1 µ a 1/n1 + 1/n a e ( (µ 1 µa)2 2 1/n 1 +1/na ) Chernoff s bound 13
93 Back to: Equalizing the Probabilities of Confusion Most simple setting: for all a {1,..., K}, ν a = N (µ a, 1). if you stop at time t, your probability of prefering arm a 2 to arm a = 1 is: P ( X a,na > X 1,n1 ) = P ( Xa,na µ a ( X1,n1 µ 1 ) 1/n1 + 1/n a > ( ) = Φ µ 1 µ a 1/n1 + 1/n a ) µ 1 µ a 1/n1 + 1/n a e ( (µ 1 µa)2 2 1/n 1 +1/na ) Large Deviation Principle = e n 1 (µ 1 m)2 2 e na(µa m)2 2 with m = n 1µ 1 + n 1 µ a n 1 + n a P ( X1,n1 < m ) P ( Xa,na m ) = max µ a<m<µ 1 P ( X1,n1 < m ) P ( Xa,na m ) (cf. Sanov) µ a m µ 1 13
94 Entropic Method for Large Deviations Lower Bounds Let d(µ, µ ) = KL ( N (µ, 1), N (µ, 1)) = (x y)2 2, KL(Y, Z) = KL ( L(Y ), L(Z) ), ɛ > 0, µ a m µ 1 and X 1,1,..., X 1,n1 iid N (µ1, 1) X a,1,..., X a,na iid N (µa, 1) X 1,1,..., X 1,n 1 iid N (m ɛ, 1) X a,1,..., X a,n a iid N (m + ɛ, 1) µ a m µ 1 n 1 d(m ɛ, µ 1 ) + n a d(m + ɛ, µ a ) = KL ( (X a,i) ) KL(P P a,i, (X a,i ),Q Q ) a,i = KL(P, Q) + KL(P, Q ) KL (1 { X a,na > X } { } 1,n1, 1 X a,na > X ) contraction of entropy 1,n1 = data-processing inequality ( = kl P ( X a,n a > X 1,n ) ( ) 1, P X a,na > X ) 1,n1 kl(p, q) = p log p + (1 p) log 1 p q 1 q P ( X a,na > X 1,n ) 1 1 log ( ) log(2) kl(p, q) p log 1 log 2 q P X a,na > X 1,n1 14
95 Entropic Method for Large Deviations Lower Bounds µ a m µ 1 n 1 d(m ɛ, µ 1 ) + n a d(m + ɛ, µ a ) = KL ( (X a,i) ) KL(P P a,i, (X a,i ),Q Q ) a,i = KL(P, Q) + KL(P, Q ) KL (1 { X a,na > X } { 1,n1, 1 Xa,na > X } ) contraction of entropy 1,n1 = data-processing inequality ( = kl P ( X a,na > X 1,n ) ( 1, P Xa,na > X ) ) 1,n1 kl(p, q) = p log p + (1 p) log 1 p q 1 q P ( X a,n a > X 1,n ) 1 1 log ( ) log(2) kl(p, q) p log 1 log 2 q P X a,na > X 1,n1 P ( ( Xa,na > X ) 1,n1 max exp n ) 1 d(m ɛ, µ 1 ) + n a d(m + ɛ, µ a ) + log(2) µ 1 m µ a 1 e (n1+na)ɛ2 /2 (µ 1 µ a+ɛ) 2 1/n = exp 1+1/n a + log(2) 2 ( 1 e ) n1µ1 + naµa m = (n1+na)ɛ2 /2 n 1 + n a 14
96 Entropic Method for Large Deviations Lower Bounds µ a m µ 1 n 1 d(m ɛ, µ 1 ) + n a d(m + ɛ, µ a ) = KL ( (X a,i) ) KL(P P a,i, (X a,i ),Q Q ) a,i = KL(P, Q) + KL(P, Q ) KL (1 { X a,na > X } { 1,n1, 1 Xa,na > X } ) contraction of entropy 1,n1 = data-processing inequality ( = kl P ( X a,na > X 1,n ) ( 1, P Xa,na > X ) ) 1,n1 kl(p, q) = p log p + (1 p) log 1 p q 1 q P ( X a,n a > X 1,n ) 1 1 log ( ) log(2) kl(p, q) p log 1 log 2 q P X a,na > X 1,n1 ( ) 1 = T w 1 d(m ɛ, µ 1 ) + w a d(m + ɛ, µ a ) log ( P Xa,na > X ) 1,n1 ( ) if you want to have P X a,na > X 1,n1 δ then you need T log(1/δ) w 1 d(m, µ 1 ) + w a d(m, µ a ) 14
97 Lower Bound
98 Lower-Bounding the Sample Complexity Let µ = (µ 1,..., µ K ) and λ = (λ 1,..., λ K ) be two elements of S. Uniform δ-pac Constraint [Kaufmann, Cappé, G. 15] If a (µ) a (λ), any δ-pac algorithm satisfies K [ E µ Na (τ δ ) ] d(µ a, λ a ) kl(δ, 1 δ). a=1 µ 4 µ 3 µ 2 m 2 µ 1 Let Alt(µ) = {λ : a (λ) a (µ)}. Take: λ 1 = m 2 ɛ λ 2 = m 2 + ɛ E µ [N 1 (τ δ )] d(µ 1, m 2 ɛ) + E µ [N 2 (τ δ ) ] d(µ 2, m 2 + ɛ) kl(δ, 1 δ) 16
99 Lower-Bounding the Sample Complexity Let µ = (µ 1,..., µ K ) and λ = (λ 1,..., λ K ) be two elements of S. Uniform δ-pac Constraint [Kaufmann, Cappé, G. 15] If a (µ) a (λ), any δ-pac algorithm satisfies K [ E µ Na (τ δ ) ] d(µ a, λ a ) kl(δ, 1 δ). a=1 µ 4 µ 3 µ 2 m 3 µ 1 Let Alt(µ) = {λ : a (λ) a (µ)}. Take: λ 1 = m 3 ɛ λ 3 = m 3 + ɛ E µ [N 1 (τ δ )] d(µ 1, m 2 ɛ) + E µ [N 2 (τ δ ) ] d(µ 2, m 2 + ɛ) kl(δ, 1 δ) E µ [N 1 (τ δ )] d(µ 1, m 3 ɛ) + E µ [N 3 (τ δ )] d(µ 3, m 3 + ɛ) kl(δ, 1 δ) 16
100 Lower-Bounding the Sample Complexity Let µ = (µ 1,..., µ K ) and λ = (λ 1,..., λ K ) be two elements of S. Uniform δ-pac Constraint [Kaufmann, Cappé, G. 15] If a (µ) a (λ), any δ-pac algorithm satisfies K [ E µ Na (τ δ ) ] d(µ a, λ a ) kl(δ, 1 δ). a=1 µ 4 µ 3 µ 2 m 4 µ 1 Let Alt(µ) = {λ : a (λ) a (µ)}. Take: λ 1 = m 4 ɛ λ 4 = m 4 + ɛ E µ [N 1 (τ δ )] d(µ 1, m 2 ɛ) + E µ [N 2 (τ δ ) ] d(µ 2, m 2 + ɛ) kl(δ, 1 δ) E µ [N 1 (τ δ )] d(µ 1, m 3 ɛ) + E µ [N 3 (τ δ )] d(µ 3, m 3 + ɛ) kl(δ, 1 δ) E µ [N 1 (τ δ )] d(µ 1, m 4 ɛ) + E µ [N 4 (τ δ )] d(µ 4, m 4 + ɛ) kl(δ, 1 δ) 16
101 Lower-Bounding the Sample Complexity Let µ = (µ 1,..., µ K ) and λ = (λ 1,..., λ K ) be two elements of S. Uniform δ-pac Constraint [Kaufmann, Cappé, G. 15] If a (µ) a (λ), any δ-pac algorithm satisfies K [ E µ Na (τ δ ) ] d(µ a, λ a ) kl(δ, 1 δ). a=1 Let Alt(µ) = {λ : a (λ) a (µ)}. µ 4 µ 3 µ 2 m 4 m 3 m 2 µ 1 E µ [τ δ ] E µ [τ δ ] inf λ Alt(µ) a=1 inf λ Alt(µ) a=1 ( sup w Σ K K E µ [N a (τ δ )] d(µ a, λ a ) kl(δ, 1 δ) K inf E µ [N a (τ δ )] E µ [τ δ ] λ Alt(µ) a=1 d(µ a, λ a ) kl(δ, 1 δ) ) K w a d(µ a, λ a ) kl(δ, 1 δ) 16
102 Lower Bound: the Complexity of BAI Theorem For any δ-pac algorithm, E µ [τ δ ] T (µ) kl(δ, 1 δ), where T (µ) 1 = sup w Σ K inf λ Alt(µ) ( K ) w a d(µ a, λ a ). a=1 kl(δ, 1 δ) log(1/δ) when δ 0, kl(δ, 1 δ) log ( 1/(2.4δ) ) cf. [Graves and Lai 1997, Vaidhyan and Sundaresan, 2015] the optimal proportions of arm draws are ( K ) w (µ) = argmax w Σ K inf λ Alt(µ) a=1 w a d(µ a, λ a ) they do not depend on δ 17
103 PAC-BAI as a Game Given a parameter µ = (µ 1,..., µ K ) : the statistician chooses proportions of arm draws w = (w a ) a the opponent chooses an alternative model λ the payoff is the minimal number T = T (w, λ) of draws necessary to ensure that he does not violate the δ-pac constraint K Tw a d(µ a, λ a ) kl(δ, 1 δ) a=1 T (µ) kl(δ, 1 δ) = value of the game w = optimal action for the statistician 18
104 PAC-BAI as a Game Given a parameter µ = (µ 1,..., µ K ) such that µ 1 > µ 2 µ K : the statistician chooses proportions of arm draws w = (w a ) a the opponent chooses an arm a {2,..., K} and λ a = arg min λ w 1 d(µ 1, λ) + w a d(µ a, λ) µ 4 µ 3 µ 2 m 3 µ 1 the payoff is the minimal number T = T (w, a, δ) of draws necessary to ensure that Tw 1 d(µ 1, λ a ɛ) + Tw a d(µ a, λ a + ɛ) kl(δ, 1 δ) kl(δ, 1 δ) that is T (w, a, δ) = w 1 d(µ 1, λ a ɛ) + w a d(µ a, λ a + ɛ) T (µ) kl(δ, 1 δ) = value of the game w = optimal action for the statistician 18
105 Properties of T (µ) and w (µ) 1. Unique solution, solution of scalar equations only 2. For all µ S, for all a, w a (µ) > 0 3. w is continuous in every µ S 4. If µ 1 > µ 2 µ K, one has w 2 (µ) w K (µ) (one may have w 1 (µ) < w 2 (µ)) 5. Case of two arms [Kaufmann, Cappé, G. 14]: E µ [τ δ ] kl(δ, 1 δ) d (µ 1, µ 2 ). where d is the reversed Chernoff information d (µ 1, µ 2 ) := d(µ 1, µ ) = d(µ 2, µ ). 6. Gaussian arms : algebraic equation but no simple formula for K 3. K 2σ 2 K 2 T 2σ 2 (µ) 2 a 2. a a=1 a=1 19
106 The Track-and-Stop Strategy
107 Sampling rule: Tracking the optimal proportions ˆµ(t) = (ˆµ 1 (t),..., ˆµ K (t)): vector of empirical means Introducing U t = { a : N a (t) < } t, the arm sampled at round t + 1 is argmin N a (t) if U t (forced exploration) a U A t+1 t argmax t wa ( ˆµ(t)) N a (t) (tracking) 1 a K Lemma Under the Tracking sampling rule, ( P µ lim t N a (t) t ) = wa (µ) = 1. 21
108 Sequential Generalized Likelihood Test High values of the Generalized Likelihood Ratio Z a,b (t) := log max {λ:λ a λ b } dp λ (X 1,..., X t ) max {λ:λa λ b } dp λ (X 1,..., X t ) = N a (t) d (ˆµ a (t), ˆµ a,b (t) ) + N b (t) d (ˆµ b (t), ˆµ a,b (t) ) if ˆµ a (t) > ˆµ b (t) Z b,a (t) otherwise reject the hypothesis that (µ a µ b ). We stop when one arm is assessed to be significantly larger than all other arms, according to a GLR test: τ δ = inf { t N : a {1,..., K}, b a, Z a,b (t) > β(t, δ) } { } = inf t N : Z(t) := max min Z a,b(t) > β(t, δ) a {1,...,K} b a Chernoff stopping rule [Chernoff 59] Two other possible interpretations of the stopping rule: MDL: Z a,b (t) = ( N a(t) + N b (t) ) H (ˆµ a,b (t) ) [ N a(t)h (ˆµ a(t) ) + N b (t)h (ˆµ b (t) )] 22
109 Sequential Generalized Likelihood Test High values of the Generalized Likelihood Ratio Z a,b (t) := log max {λ:λ a λ b } dp λ (X 1,..., X t ) max {λ:λa λ b } dp λ (X 1,..., X t ) reject the hypothesis that (µ a µ b ). We stop when one arm is assessed to be significantly larger than all other arms, according to a GLR test: { } τ δ = inf t N : Z(t) := max min Z a,b(t) > β(t, δ) a {1,...,K} b a Chernoff stopping rule [Chernoff 59] Two other possible interpretations of the stopping rule: plug-in complexity estimate: with F (w, µ) := stop when Z(t) = t F t T (µ) = t F (w, µ) kl(δ, 1 δ). inf K w d ( ) µ a, λ a, λ Alt(µ) a=1 ( Na(t) ), ˆµ(t) β(t, δ) instead of the lower bound t 22
110 Calibration Theorem The Chernoff rule is δ-pac for β(t, δ) = log ( ) 2(K 1)t δ Lemma If µ a < µ b, whatever the sampling rule, P µ ( t N : Z a,b (t) > log ( )) 2t δ δ The proof uses: Barron s lemma (change of distribution) and Krichevsky-Trofimov s universal distribution (very information-theoretic ideas) 23
111 Asymptotic Optimality of the T&S strategy Theorem The Track-and-Stop strategy, that uses the Tracking sampling rule the Chernoff stopping rule with β(t, δ) = log and recommends â τδ = argmax a=1...k ˆµ a (τ δ ) is δ-pac for every δ (0, 1) and satisfies lim sup δ 0 E µ [τ δ ] log(1/δ) = T (µ). ( ) 2(K 1)t δ 24
112 Why is the T&S Strategy asymptotically Optimal? 25
113 Why is the T&S Strategy asymptotically Optimal? 25
114 Why is the T&S Strategy asymptotically Optimal? 25
115 Why is the T&S Strategy asymptotically Optimal? 25
116 Why is the T&S Strategy asymptotically Optimal? 25
117 Why is the T&S Strategy asymptotically Optimal? 25
118 Why is the T&S Strategy asymptotically Optimal? 25
119 Why is the T&S Strategy asymptotically Optimal? 25
120 Why is the T&S Strategy asymptotically Optimal? 25
121 Why is the T&S Strategy asymptotically Optimal? 25
122 Why is the T&S Strategy asymptotically Optimal? 25
123 Why is the T&S Strategy asymptotically Optimal? 25
124 Why is the T&S Strategy asymptotically Optimal? 25
125 Why is the T&S Strategy asymptotically Optimal? 25
126 Why is the T&S Strategy asymptotically Optimal? 25
127 Why is the T&S Strategy asymptotically Optimal? 25
128 Why is the T&S Strategy asymptotically Optimal? 25
129 Why is the T&S Strategy asymptotically Optimal? 25
130 Why is the T&S Strategy asymptotically Optimal? 25
131 Why is the T&S Strategy asymptotically Optimal? 25
132 Sketch of proof (almost-sure convergence only) forced exploration = N a (t) a.s. for all a {1,..., K} µ(t) µ a.s. w ( ˆµ(t) ) w a.s. tracking rule: N a (t) t t w a but the mapping F : (µ, w) a.s. inf λ Alt(µ ) a=1 K w a d(µ a, λ a ) is continuous at ( (µ, w (µ)): ) Z(t) = t F ˆµ(t), (N a (t)/t) K a=1 t F (µ, w ) = t T (µ) 1 and for every ɛ > 0 there exists t 0 such that t t 0 Z(t) t (1 + ɛ) 1 T (µ) 1 { } = Thus τ δ t 0 inf t N : (1 + ɛ) 1 T (µ) 1 t log(2(k 1)t/δ) τ δ and lim sup δ 0 log(1/δ) (1 + ɛ)t (µ) a.s. 26
133 Numerical Experiments µ 1 = [ ] w (µ 1 ) = [ ] µ 2 = [ ] w (µ 2 ) = [ ] In practice, set the threshold to β(t, δ) = log ( ) log(t)+1 δ (δ-pac OK) Track-and-Stop Chernoff-Racing KL-LUCB KL-Racing µ µ Table 1: Expected number of draws E µ[τ δ ] for δ = 0.1, averaged over N = 3000 experiments. Empirically good even for large values of the risk δ Racing is sub-optimal in general, because it plays w 1 = w 2 LUCB is sub-optimal in general, because it plays w 1 = 1/2 27
134 Perspectives For best arm identification, we showed that inf lim sup PAC algorithm δ 0 E µ [τ δ ] log(1/δ) = sup w Σ K inf λ Alt(µ) ( K ) w a d(µ a, λ a ) and provided an efficient strategy asymptotically matching this bound. Future work: (easy) anytime stopping gives a confidence level (easy) find an ɛ-optimal arm (easy) find the m-best arms design and analyze more stable algorithm (hint: optimism) give a simple algorithm with a finite-time analysis candidate: play action maximizing the expected increase of Z(t) extend to structured and continuous settings a=1 x 1 x 2 x 3 x 4 x 5 28
135 References O. Cappé, A. Garivier, O-A. Maillard, R. Munos, and G. Stoltz. Kullback-Leibler upper confidence bounds for optimal sequential allocation. Annals of Statistics, H. Chernoff. Sequential design of Experiments. The Annals of Mathematical Statistics, E. Even-Dar, S. Mannor, Y. Mansour, Action Elimination and Stopping Conditions for the Multi-Armed Bandit and Reinforcement Learning Problems. JMLR, T.L. Graves and T.L. Lai. Asymptotically Efficient adaptive choice of control laws in controlled markov chains. SIAM Journal on Control and Optimization, 35(3):715743, S. Kalyanakrishnan, A. Tewari, P. Auer, and P. Stone. PAC subset selection in stochastic multi- armed bandits. ICML, E. Kaufmann, O. Cappé, A. Garivier. On the Complexity of Best Arm Identification in Multi-Armed Bandit Models. JMLR, 2015 A. Garivier, E. Kaufmann. Optimal Best Arm Identification with Fixed Confidence, COLT 16, New York, arxiv: A. Garivier, P. Ménard, G. Stoltz. Explore First, Exploit Next: The True Shape of Regret in Bandit Problems. E. Kaufmann, S. Kalyanakrishnan. The information complexity of best arm identification, COLT 2013 T.L. Lai and H. Robbins. Asymptotically efficient adaptive allocation rules. Advances in Applied Mathematics, D. Russo. Simple Bayesian Algorithms for Best Arm Identification, COLT 2016 N.K. Vaidhyan and R. Sundaresan. Learning to detect an oddball target. arxiv: ,
On the Complexity of Best Arm Identification with Fixed Confidence
On the Complexity of Best Arm Identification with Fixed Confidence Discrete Optimization with Noise Aurélien Garivier, joint work with Emilie Kaufmann CNRS, CRIStAL) to be presented at COLT 16, New York
More informationOn Sequential Decision Problems
On Sequential Decision Problems Aurélien Garivier Séminaire du LIP, 14 mars 2018 Équipe-projet AOC: Apprentissage, Optimisation, Complexité Institut de Mathématiques de Toulouse LabeX CIMI Université Paul
More informationThe information complexity of best-arm identification
The information complexity of best-arm identification Emilie Kaufmann, joint work with Olivier Cappé and Aurélien Garivier MAB workshop, Lancaster, January th, 206 Context: the multi-armed bandit model
More informationThe information complexity of sequential resource allocation
The information complexity of sequential resource allocation Emilie Kaufmann, joint work with Olivier Cappé, Aurélien Garivier and Shivaram Kalyanakrishan SMILE Seminar, ENS, June 8th, 205 Sequential allocation
More informationTwo optimization problems in a stochastic bandit model
Two optimization problems in a stochastic bandit model Emilie Kaufmann joint work with Olivier Cappé, Aurélien Garivier and Shivaram Kalyanakrishnan Journées MAS 204, Toulouse Outline From stochastic optimization
More informationOn the Complexity of Best Arm Identification in Multi-Armed Bandit Models
On the Complexity of Best Arm Identification in Multi-Armed Bandit Models Aurélien Garivier Institut de Mathématiques de Toulouse Information Theory, Learning and Big Data Simons Institute, Berkeley, March
More informationRevisiting the Exploration-Exploitation Tradeoff in Bandit Models
Revisiting the Exploration-Exploitation Tradeoff in Bandit Models joint work with Aurélien Garivier (IMT, Toulouse) and Tor Lattimore (University of Alberta) Workshop on Optimization and Decision-Making
More informationStratégies bayésiennes et fréquentistes dans un modèle de bandit
Stratégies bayésiennes et fréquentistes dans un modèle de bandit thèse effectuée à Telecom ParisTech, co-dirigée par Olivier Cappé, Aurélien Garivier et Rémi Munos Journées MAS, Grenoble, 30 août 2016
More informationMulti-armed bandit models: a tutorial
Multi-armed bandit models: a tutorial CERMICS seminar, March 30th, 2016 Multi-Armed Bandit model: general setting K arms: for a {1,..., K}, (X a,t ) t N is a stochastic process. (unknown distributions)
More informationOptimal Best Arm Identification with Fixed Confidence
JMLR: Workshop and Conference Proceedings vol 49:1 30, 2016 Optimal Best Arm Identification with Fixed Confidence Aurélien Garivier Institut de Mathématiques de Toulouse; UMR5219 Université de Toulouse;
More informationBandit Algorithms for Pure Exploration: Best Arm Identification and Game Tree Search. Wouter M. Koolen
Bandit Algorithms for Pure Exploration: Best Arm Identification and Game Tree Search Wouter M. Koolen Machine Learning and Statistics for Structures Friday 23 rd February, 2018 Outline 1 Intro 2 Model
More informationBayesian and Frequentist Methods in Bandit Models
Bayesian and Frequentist Methods in Bandit Models Emilie Kaufmann, Telecom ParisTech Bayes In Paris, ENSAE, October 24th, 2013 Emilie Kaufmann (Telecom ParisTech) Bayesian and Frequentist Bandits BIP,
More informationOnline Learning and Sequential Decision Making
Online Learning and Sequential Decision Making Emilie Kaufmann CNRS & CRIStAL, Inria SequeL, emilie.kaufmann@univ-lille.fr Research School, ENS Lyon, Novembre 12-13th 2018 Emilie Kaufmann Sequential Decision
More informationOn Bayesian bandit algorithms
On Bayesian bandit algorithms Emilie Kaufmann joint work with Olivier Cappé, Aurélien Garivier, Nathaniel Korda and Rémi Munos July 1st, 2012 Emilie Kaufmann (Telecom ParisTech) On Bayesian bandit algorithms
More informationarxiv: v2 [stat.ml] 14 Nov 2016
Journal of Machine Learning Research 6 06-4 Submitted 7/4; Revised /5; Published /6 On the Complexity of Best-Arm Identification in Multi-Armed Bandit Models arxiv:407.4443v [stat.ml] 4 Nov 06 Emilie Kaufmann
More informationOn the Complexity of A/B Testing
JMLR: Workshop and Conference Proceedings vol 35:1 3, 014 On the Complexity of A/B Testing Emilie Kaufmann LTCI, Télécom ParisTech & CNRS KAUFMANN@TELECOM-PARISTECH.FR Olivier Cappé CAPPE@TELECOM-PARISTECH.FR
More informationarxiv: v3 [cs.lg] 7 Nov 2017
ESAIM: PROCEEDINGS AND SURVEYS, Vol.?, 2017, 1-10 Editors: Will be set by the publisher arxiv:1702.00001v3 [cs.lg] 7 Nov 2017 LEARNING THE DISTRIBUTION WITH LARGEST MEAN: TWO BANDIT FRAMEWORKS Emilie Kaufmann
More informationBandit models: a tutorial
Gdt COS, December 3rd, 2015 Multi-Armed Bandit model: general setting K arms: for a {1,..., K}, (X a,t ) t N is a stochastic process. (unknown distributions) Bandit game: a each round t, an agent chooses
More informationBandits : optimality in exponential families
Bandits : optimality in exponential families Odalric-Ambrym Maillard IHES, January 2016 Odalric-Ambrym Maillard Bandits 1 / 40 Introduction 1 Stochastic multi-armed bandits 2 Boundary crossing probabilities
More informationRegret Analysis of Stochastic and Nonstochastic Multi-armed Bandit Problems, Part I. Sébastien Bubeck Theory Group
Regret Analysis of Stochastic and Nonstochastic Multi-armed Bandit Problems, Part I Sébastien Bubeck Theory Group i.i.d. multi-armed bandit, Robbins [1952] i.i.d. multi-armed bandit, Robbins [1952] Known
More informationInformational Confidence Bounds for Self-Normalized Averages and Applications
Informational Confidence Bounds for Self-Normalized Averages and Applications Aurélien Garivier Institut de Mathématiques de Toulouse - Université Paul Sabatier Thursday, September 12th 2013 Context Tree
More informationDynamic resource allocation: Bandit problems and extensions
Dynamic resource allocation: Bandit problems and extensions Aurélien Garivier Institut de Mathématiques de Toulouse MAD Seminar, Université Toulouse 1 October 3rd, 2014 The Bandit Model Roadmap 1 The Bandit
More informationKULLBACK-LEIBLER UPPER CONFIDENCE BOUNDS FOR OPTIMAL SEQUENTIAL ALLOCATION
Submitted to the Annals of Statistics arxiv: math.pr/0000000 KULLBACK-LEIBLER UPPER CONFIDENCE BOUNDS FOR OPTIMAL SEQUENTIAL ALLOCATION By Olivier Cappé 1, Aurélien Garivier 2, Odalric-Ambrym Maillard
More informationThe Multi-Arm Bandit Framework
The Multi-Arm Bandit Framework A. LAZARIC (SequeL Team @INRIA-Lille) ENS Cachan - Master 2 MVA SequeL INRIA Lille MVA-RL Course In This Lecture A. LAZARIC Reinforcement Learning Algorithms Oct 29th, 2013-2/94
More informationReinforcement Learning
Reinforcement Learning Lecture 5: Bandit optimisation Alexandre Proutiere, Sadegh Talebi, Jungseul Ok KTH, The Royal Institute of Technology Objectives of this lecture Introduce bandit optimisation: the
More informationarxiv: v2 [math.st] 14 Nov 2016
On Explore-hen-Commit Strategies arxiv:1605.08988v math.s] 1 Nov 016 Aurélien Garivier Institut de Mathématiques de oulouse; UMR519 Université de oulouse; CNRS UPS IM, F-3106 oulouse Cedex 9, France aurelien.garivier@math.univ-toulouse.fr
More informationIntroduction to Bandit Algorithms. Introduction to Bandit Algorithms
Stochastic K-Arm Bandit Problem Formulation Consider K arms (actions) each correspond to an unknown distribution {ν k } K k=1 with values bounded in [0, 1]. At each time t, the agent pulls an arm I t {1,...,
More informationSequential Test for the Lowest Mean: From Thompson to Murphy Sampling
Sequential Test for the Lowest Mean: From Thompson to Murphy Sampling Emilie Kaufmann 1 Wouter M. Koolen 2 Aurélien Garivier 3 1 CNRS & U. Lille, CRIStAL / SequeL Inria Lille, emilie.kaufmann@univ-lille.fr
More informationTwo generic principles in modern bandits: the optimistic principle and Thompson sampling
Two generic principles in modern bandits: the optimistic principle and Thompson sampling Rémi Munos INRIA Lille, France CSML Lunch Seminars, September 12, 2014 Outline Two principles: The optimistic principle
More informationA minimax and asymptotically optimal algorithm for stochastic bandits
Proceedings of Machine Learning Research 76:1 15, 017 Algorithmic Learning heory 017 A minimax and asymptotically optimal algorithm for stochastic bandits Pierre Ménard Aurélien Garivier Institut de Mathématiques
More informationRacing Thompson: an Efficient Algorithm for Thompson Sampling with Non-conjugate Priors
Racing Thompson: an Efficient Algorithm for Thompson Sampling with Non-conugate Priors Yichi Zhou 1 Jun Zhu 1 Jingwe Zhuo 1 Abstract Thompson sampling has impressive empirical performance for many multi-armed
More informationSubsampling, Concentration and Multi-armed bandits
Subsampling, Concentration and Multi-armed bandits Odalric-Ambrym Maillard, R. Bardenet, S. Mannor, A. Baransi, N. Galichet, J. Pineau, A. Durand Toulouse, November 09, 2015 O-A. Maillard Subsampling and
More informationMonte-Carlo Tree Search by. MCTS by Best Arm Identification
Monte-Carlo Tree Search by Best Arm Identification and Wouter M. Koolen Inria Lille SequeL team CWI Machine Learning Group Inria-CWI workshop Amsterdam, September 20th, 2017 Part of...... a new Associate
More informationarxiv: v2 [stat.ml] 19 Jul 2012
Thompson Sampling: An Asymptotically Optimal Finite Time Analysis Emilie Kaufmann, Nathaniel Korda and Rémi Munos arxiv:105.417v [stat.ml] 19 Jul 01 Telecom Paristech UMR CNRS 5141 & INRIA Lille - Nord
More informationThe multi armed-bandit problem
The multi armed-bandit problem (with covariates if we have time) Vianney Perchet & Philippe Rigollet LPMA Université Paris Diderot ORFE Princeton University Algorithms and Dynamics for Games and Optimization
More informationAdvanced Machine Learning
Advanced Machine Learning Bandit Problems MEHRYAR MOHRI MOHRI@ COURANT INSTITUTE & GOOGLE RESEARCH. Multi-Armed Bandit Problem Problem: which arm of a K-slot machine should a gambler pull to maximize his
More informationStat 260/CS Learning in Sequential Decision Problems.
Stat 260/CS 294-102. Learning in Sequential Decision Problems. Peter Bartlett 1. Multi-armed bandit algorithms. Exponential families. Cumulant generating function. KL-divergence. KL-UCB for an exponential
More informationLecture 4: Lower Bounds (ending); Thompson Sampling
CMSC 858G: Bandits, Experts and Games 09/12/16 Lecture 4: Lower Bounds (ending); Thompson Sampling Instructor: Alex Slivkins Scribed by: Guowei Sun,Cheng Jie 1 Lower bounds on regret (ending) Recap from
More informationKullback-Leibler Upper Confidence Bounds for Optimal Sequential Allocation
Kullback-Leibler Upper Confidence Bounds for Optimal Sequential Allocation Olivier Cappé, Aurélien Garivier, Odalric-Ambrym Maillard, Rémi Munos, Gilles Stoltz To cite this version: Olivier Cappé, Aurélien
More informationBandit Algorithms. Tor Lattimore & Csaba Szepesvári
Bandit Algorithms Tor Lattimore & Csaba Szepesvári Bandits Time 1 2 3 4 5 6 7 8 9 10 11 12 Left arm $1 $0 $1 $1 $0 Right arm $1 $0 Five rounds to go. Which arm would you play next? Overview What are bandits,
More informationStat 260/CS Learning in Sequential Decision Problems. Peter Bartlett
Stat 260/CS 294-102. Learning in Sequential Decision Problems. Peter Bartlett 1. Multi-armed bandit algorithms. Concentration inequalities. P(X ǫ) exp( ψ (ǫ))). Cumulant generating function bounds. Hoeffding
More informationOptimality of Thompson Sampling for Gaussian Bandits Depends on Priors
Optimality of Thompson Sampling for Gaussian Bandits Depends on Priors Junya Honda Akimichi Takemura The University of Tokyo {honda, takemura}@stat.t.u-tokyo.ac.jp Abstract In stochastic bandit problems,
More informationPure Exploration in Infinitely-Armed Bandit Models with Fixed-Confidence
Pure Exploration in Infinitely-Armed Bandit Models with Fixed-Confidence Maryam Aziz, Jesse Anderton, Emilie Kaufmann, Javed Aslam To cite this version: Maryam Aziz, Jesse Anderton, Emilie Kaufmann, Javed
More informationarxiv: v1 [stat.ml] 27 May 2016
An optimal algorithm for the hresholding Bandit Problem Andrea Locatelli Maurilio Gutzeit Alexandra Carpentier Department of Mathematics, University of Potsdam, Germany ANDREA.LOCAELLI@UNI-POSDAM.DE MGUZEI@UNI-POSDAM.DE
More informationNearly Optimal Sampling Algorithms for Combinatorial Pure Exploration
JMLR: Workshop and Conference Proceedings vol 65: 55, 207 30th Annual Conference on Learning Theory Nearly Optimal Sampling Algorithms for Combinatorial Pure Exploration Editor: Under Review for COLT 207
More informationOrdinal optimization - Empirical large deviations rate estimators, and multi-armed bandit methods
Ordinal optimization - Empirical large deviations rate estimators, and multi-armed bandit methods Sandeep Juneja Tata Institute of Fundamental Research Mumbai, India joint work with Peter Glynn Applied
More informationLecture 5: Regret Bounds for Thompson Sampling
CMSC 858G: Bandits, Experts and Games 09/2/6 Lecture 5: Regret Bounds for Thompson Sampling Instructor: Alex Slivkins Scribed by: Yancy Liao Regret Bounds for Thompson Sampling For each round t, we defined
More informationarxiv: v1 [cs.ds] 4 Mar 2016
Sequential ranking under random semi-bandit feedback arxiv:1603.01450v1 [cs.ds] 4 Mar 2016 Hossein Vahabi, Paul Lagrée, Claire Vernade, Olivier Cappé March 7, 2016 Abstract In many web applications, a
More informationRobustness of Anytime Bandit Policies
Robustness of Anytime Bandit Policies Antoine Salomon, Jean-Yves Audibert To cite this version: Antoine Salomon, Jean-Yves Audibert. Robustness of Anytime Bandit Policies. 011. HAL Id:
More informationCorrupt Bandits. Abstract
Corrupt Bandits Pratik Gajane Orange labs/inria SequeL Tanguy Urvoy Orange labs Emilie Kaufmann INRIA SequeL pratik.gajane@inria.fr tanguy.urvoy@orange.com emilie.kaufmann@inria.fr Editor: Abstract We
More informationAdaptive Concentration Inequalities for Sequential Decision Problems
Adaptive Concentration Inequalities for Sequential Decision Problems Shengjia Zhao Tsinghua University zhaosj12@stanford.edu Ashish Sabharwal Allen Institute for AI AshishS@allenai.org Enze Zhou Tsinghua
More informationarxiv: v4 [math.pr] 26 Aug 2013
The Annals of Statistics 2013, Vol. 41, No. 3, 1516 1541 DOI: 10.1214/13-AOS1119 c Institute of Mathematical Statistics, 2013 arxiv:1210.1136v4 [math.pr] 26 Aug 2013 KULLBACK LEIBLER UPPER CONFIDENCE BOUNDS
More informationMultiple Identifications in Multi-Armed Bandits
Multiple Identifications in Multi-Armed Bandits arxiv:05.38v [cs.lg] 4 May 0 Sébastien Bubeck Department of Operations Research and Financial Engineering, Princeton University sbubeck@princeton.edu Tengyao
More informationMulti armed bandit problem: some insights
Multi armed bandit problem: some insights July 4, 20 Introduction Multi Armed Bandit problems have been widely studied in the context of sequential analysis. The application areas include clinical trials,
More informationTHE MULTI-ARMED BANDIT PROBLEM: AN EFFICIENT NON-PARAMETRIC SOLUTION
THE MULTI-ARMED BANDIT PROBLEM: AN EFFICIENT NON-PARAMETRIC SOLUTION Hock Peng Chan stachp@nus.edu.sg Department of Statistics and Applied Probability National University of Singapore Abstract Lai and
More informationOrdinal Optimization and Multi Armed Bandit Techniques
Ordinal Optimization and Multi Armed Bandit Techniques Sandeep Juneja. with Peter Glynn September 10, 2014 The ordinal optimization problem Determining the best of d alternative designs for a system, on
More informationPure Exploration Stochastic Multi-armed Bandits
C&A Workshop 2016, Hangzhou Pure Exploration Stochastic Multi-armed Bandits Jian Li Institute for Interdisciplinary Information Sciences Tsinghua University Outline Introduction 2 Arms Best Arm Identification
More informationStochastic bandits: Explore-First and UCB
CSE599s, Spring 2014, Online Learning Lecture 15-2/19/2014 Stochastic bandits: Explore-First and UCB Lecturer: Brendan McMahan or Ofer Dekel Scribe: Javad Hosseini In this lecture, we like to answer this
More informationAnytime optimal algorithms in stochastic multi-armed bandits
Rémy Degenne LPMA, Université Paris Diderot Vianney Perchet CREST, ENSAE REMYDEGENNE@MATHUNIV-PARIS-DIDEROTFR VIANNEYPERCHET@NORMALESUPORG Abstract We introduce an anytime algorithm for stochastic multi-armed
More informationBest Arm Identification in Multi-Armed Bandits
Best Arm Identification in Multi-Armed Bandits Jean-Yves Audibert Imagine, Université Paris Est & Willow, CNRS/ENS/INRIA, Paris, France audibert@certisenpcfr Sébastien Bubeck, Rémi Munos SequeL Project,
More informationBandits and Exploration: How do we (optimally) gather information? Sham M. Kakade
Bandits and Exploration: How do we (optimally) gather information? Sham M. Kakade Machine Learning for Big Data CSE547/STAT548 University of Washington S. M. Kakade (UW) Optimization for Big data 1 / 22
More informationOnline Learning with Gaussian Payoffs and Side Observations
Online Learning with Gaussian Payoffs and Side Observations Yifan Wu 1 András György 2 Csaba Szepesvári 1 1 Department of Computing Science University of Alberta 2 Department of Electrical and Electronic
More informationLearning to play K-armed bandit problems
Learning to play K-armed bandit problems Francis Maes 1, Louis Wehenkel 1 and Damien Ernst 1 1 University of Liège Dept. of Electrical Engineering and Computer Science Institut Montefiore, B28, B-4000,
More informationIntroduction to Reinforcement Learning Part 3: Exploration for decision making, Application to games, optimization, and planning
Introduction to Reinforcement Learning Part 3: Exploration for decision making, Application to games, optimization, and planning Rémi Munos SequeL project: Sequential Learning http://researchers.lille.inria.fr/
More informationAdaptive Concentration Inequalities for Sequential Decision Problems
Adaptive Concentration Inequalities for Sequential Decision Problems Shengjia Zhao Tsinghua University zhaosj12@stanford.edu Ashish Sabharwal Allen Institute for AI AshishS@allenai.org Enze Zhou Tsinghua
More informationIntroduction to Statistical Learning Theory
Introduction to Statistical Learning Theory In the last unit we looked at regularization - adding a w 2 penalty. We add a bias - we prefer classifiers with low norm. How to incorporate more complicated
More informationarxiv: v1 [cs.lg] 14 Nov 2018 Abstract
Sample complexity of partition identification using multi-armed bandits Sandeep Juneja Subhashini Krishnasamy TIFR, Mumbai November 15, 2018 arxiv:1811.05654v1 [cs.lg] 14 Nov 2018 Abstract Given a vector
More informationThompson Sampling for the non-stationary Corrupt Multi-Armed Bandit
European Worshop on Reinforcement Learning 14 (2018 October 2018, Lille, France. Thompson Sampling for the non-stationary Corrupt Multi-Armed Bandit Réda Alami Orange Labs 2 Avenue Pierre Marzin 22300,
More informationBoundary Crossing for General Exponential Families
Proceedings of Machine Learning Research 76:1 34, 017 Algorithmic Learning Theory 017 Boundary Crossing for General Exponential Families Odalric-Ambrym Maillard INRIA Lille - Nord Europe, Villeneuve d
More informationarxiv: v3 [stat.ml] 6 Nov 2017
On Bayesian index policies for sequential resource allocation Emilie Kaufmann arxiv:60.090v3 [stat.ml] 6 Nov 07 CNRS & Univ. Lille, UMR 989 CRIStAL, Inria SequeL emilie.kaufmann@univ-lille.fr Abstract
More informationInformation Theory and Hypothesis Testing
Summer School on Game Theory and Telecommunications Campione, 7-12 September, 2014 Information Theory and Hypothesis Testing Mauro Barni University of Siena September 8 Review of some basic results linking
More informationPAC Subset Selection in Stochastic Multi-armed Bandits
In Langford, Pineau, editors, Proceedings of the 9th International Conference on Machine Learning, pp 655--66, Omnipress, New York, NY, USA, 0 PAC Subset Selection in Stochastic Multi-armed Bandits Shivaram
More informationCsaba Szepesvári 1. University of Alberta. Machine Learning Summer School, Ile de Re, France, 2008
LEARNING THEORY OF OPTIMAL DECISION MAKING PART I: ON-LINE LEARNING IN STOCHASTIC ENVIRONMENTS Csaba Szepesvári 1 1 Department of Computing Science University of Alberta Machine Learning Summer School,
More informationPure Exploration Stochastic Multi-armed Bandits
CAS2016 Pure Exploration Stochastic Multi-armed Bandits Jian Li Institute for Interdisciplinary Information Sciences Tsinghua University Outline Introduction Optimal PAC Algorithm (Best-Arm, Best-k-Arm):
More informationRegret Analysis of Stochastic and Nonstochastic Multi-armed Bandit Problems. Sébastien Bubeck Theory Group
Regret Analysis of Stochastic and Nonstochastic Multi-armed Bandit Problems Sébastien Bubeck Theory Group Part 1: i.i.d., adversarial, and Bayesian bandit models i.i.d. multi-armed bandit, Robbins [1952]
More informationThe knowledge gradient method for multi-armed bandit problems
The knowledge gradient method for multi-armed bandit problems Moving beyond inde policies Ilya O. Ryzhov Warren Powell Peter Frazier Department of Operations Research and Financial Engineering Princeton
More informationConsistency of the maximum likelihood estimator for general hidden Markov models
Consistency of the maximum likelihood estimator for general hidden Markov models Jimmy Olsson Centre for Mathematical Sciences Lund University Nordstat 2012 Umeå, Sweden Collaborators Hidden Markov models
More informationStatistical Inference
Statistical Inference Robert L. Wolpert Institute of Statistics and Decision Sciences Duke University, Durham, NC, USA Week 12. Testing and Kullback-Leibler Divergence 1. Likelihood Ratios Let 1, 2, 2,...
More information1 MDP Value Iteration Algorithm
CS 0. - Active Learning Problem Set Handed out: 4 Jan 009 Due: 9 Jan 009 MDP Value Iteration Algorithm. Implement the value iteration algorithm given in the lecture. That is, solve Bellman s equation using
More informationLecture 19: UCB Algorithm and Adversarial Bandit Problem. Announcements Review on stochastic multi-armed bandit problem
Lecture 9: UCB Algorithm and Adversarial Bandit Problem EECS598: Prediction and Learning: It s Only a Game Fall 03 Lecture 9: UCB Algorithm and Adversarial Bandit Problem Prof. Jacob Abernethy Scribe:
More informationBandit Algorithms. Zhifeng Wang ... Department of Statistics Florida State University
Bandit Algorithms Zhifeng Wang Department of Statistics Florida State University Outline Multi-Armed Bandits (MAB) Exploration-First Epsilon-Greedy Softmax UCB Thompson Sampling Adversarial Bandits Exp3
More informationAnalysis of Thompson Sampling for the multi-armed bandit problem
Analysis of Thompson Sampling for the multi-armed bandit problem Shipra Agrawal Microsoft Research India shipra@microsoft.com avin Goyal Microsoft Research India navingo@microsoft.com Abstract We show
More informationCovariance function estimation in Gaussian process regression
Covariance function estimation in Gaussian process regression François Bachoc Department of Statistics and Operations Research, University of Vienna WU Research Seminar - May 2015 François Bachoc Gaussian
More informationTHE first formalization of the multi-armed bandit problem
EDIC RESEARCH PROPOSAL 1 Multi-armed Bandits in a Network Farnood Salehi I&C, EPFL Abstract The multi-armed bandit problem is a sequential decision problem in which we have several options (arms). We can
More informationGeneralized Neyman Pearson optimality of empirical likelihood for testing parameter hypotheses
Ann Inst Stat Math (2009) 61:773 787 DOI 10.1007/s10463-008-0172-6 Generalized Neyman Pearson optimality of empirical likelihood for testing parameter hypotheses Taisuke Otsu Received: 1 June 2007 / Revised:
More informationStochastic Regret Minimization via Thompson Sampling
JMLR: Workshop and Conference Proceedings vol 35:1 22, 2014 Stochastic Regret Minimization via Thompson Sampling Sudipto Guha Department of Computer and Information Sciences, University of Pennsylvania,
More informationThe Multi-Armed Bandit Problem
The Multi-Armed Bandit Problem Electrical and Computer Engineering December 7, 2013 Outline 1 2 Mathematical 3 Algorithm Upper Confidence Bound Algorithm A/B Testing Exploration vs. Exploitation Scientist
More informationReward Maximization Under Uncertainty: Leveraging Side-Observations on Networks
Reward Maximization Under Uncertainty: Leveraging Side-Observations Reward Maximization Under Uncertainty: Leveraging Side-Observations on Networks Swapna Buccapatnam AT&T Labs Research, Middletown, NJ
More informationIntroduction to Reinforcement Learning Part 3: Exploration for sequential decision making
Introduction to Reinforcement Learning Part 3: Exploration for sequential decision making Rémi Munos SequeL project: Sequential Learning http://researchers.lille.inria.fr/ munos/ INRIA Lille - Nord Europe
More informationImproved Algorithms for Linear Stochastic Bandits
Improved Algorithms for Linear Stochastic Bandits Yasin Abbasi-Yadkori abbasiya@ualberta.ca Dept. of Computing Science University of Alberta Dávid Pál dpal@google.com Dept. of Computing Science University
More informationGeneralization bounds
Advanced Course in Machine Learning pring 200 Generalization bounds Handouts are jointly prepared by hie Mannor and hai halev-hwartz he problem of characterizing learnability is the most basic question
More informationComplexity of stochastic branch and bound methods for belief tree search in Bayesian reinforcement learning
Complexity of stochastic branch and bound methods for belief tree search in Bayesian reinforcement learning Christos Dimitrakakis Informatics Institute, University of Amsterdam, Amsterdam, The Netherlands
More informationIterative Markov Chain Monte Carlo Computation of Reference Priors and Minimax Risk
Iterative Markov Chain Monte Carlo Computation of Reference Priors and Minimax Risk John Lafferty School of Computer Science Carnegie Mellon University Pittsburgh, PA 15213 lafferty@cs.cmu.edu Abstract
More informationBandit View on Continuous Stochastic Optimization
Bandit View on Continuous Stochastic Optimization Sébastien Bubeck 1 joint work with Rémi Munos 1 & Gilles Stoltz 2 & Csaba Szepesvari 3 1 INRIA Lille, SequeL team 2 CNRS/ENS/HEC 3 University of Alberta
More informationMinimax strategy for Stratified Sampling for Monte Carlo
Journal of Machine Learning Research 1 2000 1-48 Submitted 4/00; Published 10/00 Minimax strategy for Stratified Sampling for Monte Carlo Alexandra Carpentier INRIA Lille - Nord Europe 40, avenue Halley
More informationPure Exploration in Finitely Armed and Continuous Armed Bandits
Pure Exploration in Finitely Armed and Continuous Armed Bandits Sébastien Bubeck INRIA Lille Nord Europe, SequeL project, 40 avenue Halley, 59650 Villeneuve d Ascq, France Rémi Munos INRIA Lille Nord Europe,
More informationIntroduction to Reinforcement Learning and multi-armed bandits
Introduction to Reinforcement Learning and multi-armed bandits Rémi Munos INRIA Lille - Nord Europe Currently on leave at MSR-NE http://researchers.lille.inria.fr/ munos/ NETADIS Summer School 2013, Hillerod,
More informationThe geometry of Gaussian processes and Bayesian optimization. Contal CMLA, ENS Cachan
The geometry of Gaussian processes and Bayesian optimization. Contal CMLA, ENS Cachan Background: Global Optimization and Gaussian Processes The Geometry of Gaussian Processes and the Chaining Trick Algorithm
More informationLearning to Optimize via Information-Directed Sampling
Learning to Optimize via Information-Directed Sampling Daniel Russo Stanford University Stanford, CA 94305 djrusso@stanford.edu Benjamin Van Roy Stanford University Stanford, CA 94305 bvr@stanford.edu
More information10-704: Information Processing and Learning Fall Lecture 24: Dec 7
0-704: Information Processing and Learning Fall 206 Lecturer: Aarti Singh Lecture 24: Dec 7 Note: These notes are based on scribed notes from Spring5 offering of this course. LaTeX template courtesy of
More information