Tweet
Tweet

Slide 1

Slide 1 text

ఆنͱίϯύεͱ ChainerRL Chainer Meetup #07, 9th Jun 2018 horiem@yellowshippo

Slide 2

Slide 2 text

ChainerRL Ͱ

Slide 3

Slide 3 text

࡞ਤ໰୊Λղ͖͍ͨ

Slide 4

Slide 4 text

࡞ਤ໰୊ • ఆنͱίϯύε͚ͩΛ࢖ͬͯ໨తͷਤܗΛඳ͘໰୊ http://mathworld.wolfram.com/GeometricConstruction.html

Slide 5

Slide 5 text

σϞ

Slide 6

Slide 6 text

ͷલʹ

Slide 7

Slide 7 text

ਤͷݟํ ֶशϞσϧʹ౉͢৘ใ ʢObservationʣ ਓؒ༻ ໨తͷਤܗ ར༻Մೳͳ ఺

Slide 8

Slide 8 text

σϞ

Slide 9

Slide 9 text

શମ૾ ؀ڥ ΤʔδΣϯτ ߦಈ ؍ଌ

Slide 10

Slide 10 text

શମ૾ ؀ڥ ΤʔδΣϯτ [p0_x, p0_y] [p1_x, p1_y] …… ը૾ ఺ͷ৘ใ [shape_flag, pi, pj] ৽͍͠ਤܗ

Slide 11

Slide 11 text

ωοτϫʔΫΞʔΩςΫνϟ Conv MLP MLP Conv MLP [p0_x, p0_y] [p1_x, p1_y] …… ը૾ ఺ͷ৘ใ [shape_flag, pi, pj] ৽͍͠ਤܗ

Slide 12

Slide 12 text

ωοτϫʔΫΞʔΩςΫνϟ (100, 100) (12, 3) Conv MLP MLP Conv MLP (2, 12, 12) [p0_x, p0_y] [p1_x, p1_y] …… ը૾ ఺ͷ৘ใ [shape_flag, pi, pj] ৽͍͠ਤܗ

Slide 13

Slide 13 text

ωοτϫʔΫΞʔΩςΫνϟ (100, 100) (12, 3) Conv MLP MLP Conv MLP (2, 12, 12) [p0_x, p0_y] [p1_x, p1_y] …… ը૾ ఺ͷ৘ใ [shape_flag, pi, pj] ৽͍͠ਤܗ = 288

Slide 14

Slide 14 text

ࢥͬͨ͜ͱͳͲ • ڧԽֶश΍ͬͨ͜ͱͳ͔͚ͬͨͲָ͍͠ • ChainerRL ϥΫͰΑ͍ • ߦಈۭ͕ؒେ͖͍ͷͰݮΒ͍ͨ͠ • AlphaGO ͕ࢀߟʹͳΔ͔΋ʁ • ίʔυ͖Ε͍ʹͨ͠Βެ։ && ղઆ͠·͢ • n ࣍ํఔࣜΛ ChainerRL Ͱղ͚Δ͔ʁ • ՝֎׆ಈ޷͖ͳਓɺҰॹʹ΍Γ·͠ΐ͏ʂ