Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
SantaFeTrail
Search
Sponsored
·
SiteGround - Reliable hosting with speed, security, and support you can count on.
→
TAKAMASA Saichi
October 10, 2014
Programming
150
2
Share
Embed
Copy iframe code
Copy JS code
Copy link
Start on current slide
SantaFeTrail
esysLTかどこかで発表した遺伝的プログラミングに関するスライド
TAKAMASA Saichi
October 10, 2014
More Decks by TAKAMASA Saichi
See All by TAKAMASA Saichi
メトリクス可視化から始める Rails ウェブサーバーのチューニング- Kaigi On Rails 2021
s4ichi
3
2k
Cookpad 5day service dev internship 2017 teamC
s4ichi
0
920
Region based memory management for mini ML
s4ichi
2
220
主専攻実験(春) 巡回セールスマン問題
s4ichi
0
860
Other Decks in Programming
See All in Programming
わからない話を追いかけたら、プログラミング言語を作る側にいた
ydah
3
540
Dockerfile CMD for Node.js
grazie1999
0
120
Pythonの実行はどこまで賢くなったのか? CPythonとPyPyから見る最適化のしくみ
curekoshimizu
3
1.8k
Go を使い始めて 2 ヶ月の学び / My first two months with Go
contour_gara
0
370
使いながら育てる Claude Code — 開発フローの1コマンド化 × 繰り返し指摘の自動仕組み化
shiki_kakaku
1
1.9k
S3 を使うアプリケーションをローカル完結で動かすことに全力を注いでみた / Running S3 Apps Offline
contour_gara
0
580
まずはプロンプトガイドを読もう、話はそれからだ
kiakiraki
1
200
承認済みなのに差戻しできてしまうバグ、型で潰せます
shinchit
0
100
初めての模倣学習とVLA
natsutan
0
200
引き算の組織 ― アウトカムとAIに全振りするために辞めたこと ― / Organization by Subtraction
hirokiyamamoto14
PRO
0
290
進化を続けるGo toolsの現在地 / The Current State of Ever-Evolving Go Tools
hond0413
0
250
PyConJP2026_wat_Python × Signal Processing: How to Draw Pictures with Sound Using Spectrogram Art
wat
0
290
Featured
See All Featured
Practical Orchestrator
shlominoach
191
12k
Getting science done with accelerated Python computing platforms
jacobtomlinson
2
450
SEOcharity - Dark patterns in SEO and UX: How to avoid them and build a more ethical web
sarafernandez
0
250
Art, The Web, and Tiny UX
lynnandtonic
304
22k
Learning to Love Humans: Emotional Interface Design
aarron
275
41k
Stewardship and Sustainability of Urban and Community Forests
pwiseman
0
480
JAMstack: Web Apps at Ludicrous Speed - All Things Open 2022
reverentgeek
1
580
Gemini Prompt Engineering: Practical Techniques for Tangible AI Outcomes
mfonobong
2
490
The Illustrated Children's Guide to Kubernetes
chrisshort
51
53k
Design and Strategy: How to Deal with People Who Don’t "Get" Design
morganepeng
133
19k
Design of three-dimensional binary manipulators for pick-and-place task avoiding obstacles (IECON2024)
konakalab
0
550
Bridging the Design Gap: How Collaborative Modelling removes blockers to flow between stakeholders and teams @FastFlow conf
baasie
0
660
Transcript
←͍ͭ͜ͷ coins14 @everysick
ඈͼೖΓͰ͢
ඈͼೖΓͰ͢ ਅ໘ͳ͠·͢
Ҩతϓϩάϥϛϯά
ҧ͍ ίϯϐϡʔλαΠΤϯεدΓ จܥͰΘ͔ΔΑ͏ʹؤுΓ·͢
GP(Ҩతϓϩάϥϛϯά) ͬͯͳΜͧ
GP(Ҩతϓϩάϥϛϯά) ͬͯͳΜͧ AIͬΆ͍͚Ͳҧ͏Կ͔ ϓϩάϥϜʹ੍ͱ͔ΞϧΰϦζϜΛ༩͑ͯ ੜͷਐԽͬΆ͘γϛϡϨʔτ͢Δͷ
(‘ω’)ɻ˓(???????????????????????)
GP(Ҩతϓϩάϥϛϯά) ͬͯͳΜͧ ʮ͜Μͳ෩ʹਐԽͯ͘͠ΕཔΉʯ ϓϩάϥϜʮ͑͑Ͱ͆ʯ
࣮ͷ
ࠜ અ ࢬ ༿ ߏ ྫ
։࢝ ऴྃ ॳظੈੜ ద߹ܭࢉ ऴྃ ఆ ੈҾ͖ܧ͗ ݁Ռදࣔ ॳظੈͷߏ ཚΛͬͯੜ
ަ伹ɾಥવมҟɾબ ΛϓϩάϥϜͰදݱ͠ ͦͷॲཧΛՃ͑Δ ٻΊͨղ͕ͲΕ͚ͩ ۙࣅ͍ͯ͠Δ͔ܭࢉ
Santa Fe Trail ٜͷਐԽΛγϛϡϨʔτ N*NͷϚεʹٜͱ㕒͕ஔ औಘͨ͠㕒͕ద߹ ٜ 㕒
Santa Fe Trail ٜ400ͷΤωϧΪʔΛ࣋ͭ ٜ͔Βݟͯલํʹ͔͠ਐΊͳ͍ ࠨӈʹ90ํస͢Δ͜ͱ͕Ͱ͖Δ Ҡಈ,ճసʹ͖ͭΤωϧΪʔΛ1ফඅ͢Δ
1 – Santa Fe TrailͷϓϩάϥϜ࡞ • બΞϧΰϦζϜ • બग़ͨ͠ݸମΛมԽΛՃ͑ͣ࣍ੈʹՃ •
ݸମΛͦͷ··Ҩͤ͞Δํ๏ • ࠓճͷنଇ༰ • ֤ݸମΛۉʹҨ͍ͤͨ͞ બ͢ΔݸମΛཚͰܾఆ
1 – Santa Fe TrailͷϓϩάϥϜ࡞ • ަ伹ΞϧΰϦζϜ • ҙͷ2ݸମΛΈ߹Θͤɺ৽ͨͳݸମΛੜ •
ߏͷઅಉ࢜Λަ͢ΔͳͲͷख๏ • Ұަ伹ΞϧΰϦζϜ ݸମA ݸମB
1 – Santa Fe TrailͷϓϩάϥϜ࡞ • ಥવมҟΞϧΰϦζϜ • ҙͷ1ݸମΛબग़͢Δ •
ݸମʹϥϯμϜͰੜͨ͠ߏΛΈࠐΉ • ΞϧΰϦζϜ֓ཁ ϥϯμϜͰ࡞ΒΕͨߏ બग़͞Εͨݸମ
2 -ϓϩάϥϜΛ࣮ͬͯࡍʹ݁ՌΛܭଌ 㕒ɹɹ ɹ89ݸ ελʔτɹɹࠨ্ ΤωϧΪʔɹ400
2 -ϓϩάϥϜΛ࣮ͬͯࡍʹ݁ՌΛܭଌ ϓϩάϥϜॳظঢ়ଶͰͷܭଌ݁Ռ 0 22.5 45 67.5 90 0 10
20 30 40 50 60 70 80 90 100 औಘͨ͠㕒 ੈ
͑ɺେͨ͜͠ͱͳ͘Ͷʁ
2 -ϓϩάϥϜΛ࣮ͬͯࡍʹ݁ՌΛܭଌ ϓϩάϥϜॳظঢ়ଶͰͷܭଌ݁Ռ 0 22.5 45 67.5 90 0 10
20 30 40 50 60 70 80 90 100 औಘͨ͠㕒 ੈ
3 -Ҿ͖ܧ͗ΞϧΰϦζϜΛมߋ݁͠ՌͷมԽΛܭଌ • બ • ࠷ྑબΞϧΰϦζϜ • ੈத࠷ద߹ͷߴ͍ݸମΛ݁Ռ • ղ͕ภΓ͕ͪ
• ΤϦʔτબΞϧΰϦζϜ • ݸମΛద߹ॱͰιʔτ্͠ҐNݸબͿ • Nݸ͔ΒϥϯμϜͰબͨ͠ݸମΛ݁Ռ
• ద߹ܭࢉ • ࠓ·ͰͷΞϧΰϦζϜ • औಘͨ͠㕒ͷΛద߹ • ৽ΞϧΰϦζϜ • جຊऔಘͨ͠ΞΠςϜ
• ແବͳಈ͖͝ͱʹجຊ͔ΒఆϚΠφε • ࠓճ2ͭϚΠφεΛઃஔ 3 -Ҿ͖ܧ͗ΞϧΰϦζϜΛมߋ݁͠ՌͷมԽΛܭଌ
3 -Ҿ͖ܧ͗ΞϧΰϦζϜΛมߋ݁͠ՌͷมԽΛܭଌ ΞϧΰϦζϜվྑޙͷܭଌ݁Ռ औಘͨ͠㕒 0 22.5 45 67.5 90 ੈ
0 10 20 30 40 50 60 70 80 90 100
៉ྷʹͳͬͨ✌('ω'✌ )
3 -Ҿ͖ܧ͗ΞϧΰϦζϜΛมߋ݁͠ՌͷมԽΛܭଌ ΞϧΰϦζϜվྑޙͷܭଌ݁Ռ औಘͨ͠㕒 0 22.5 45 67.5 90 ੈ
0 10 20 30 40 50 60 70 80 90 100
4 -ύϥϝʔλΛௐ͠࠷ྑͷݸମΛબग़ 1. ੈ 2. ݸମ 3. ަ伹ཱ֬ 4. ಥવมҟཱ֬
5. બཱ֬ 6. બΞϧΰϦζϜ 500[ੈ] 100[ඖ] 89% 10% 1% ΤϦʔτબ
4 -ύϥϝʔλΛௐ͠࠷ྑͷݸମΛબग़ ࠷ऴܭଌ݁Ռ औಘͨ͠㕒 0 22.5 45 67.5 90 ੈ
0 50 100 150 200 250 300 350 400 450 500
தʹશճऩͱ͔
࣮ࡍָ͍͠ ࣮͍ܰ(300ߦఔ)
͓ΘΓ