Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Sign up for free
Menu
Search
Features
All features
Private URLs
Password Protection
Custom URLS
Scheduled publishing
Remove Branding
Restrict embedding
Deck Collections
Notes
Features
All features
Private URLs
Password Protection
Custom URLS
Scheduled publishing
Remove Branding
Restrict embedding
Deck Collections
Notes
Explore
Featured decks
Featured speakers
Programming
Technology
Storyboards
Explore
Featured decks
Featured speakers
Programming
Technology
Storyboards
Pricing
Search
Sign in
Sign up for free
TGS Salt Identification Challenge 12th place so...
Search
catla
November 30, 2018
Research
12k
3
Share
Embed
Copy iframe code
Copy JS code
Copy link
Start on current slide
TGS Salt Identification Challenge 12th place solution
catla
November 30, 2018
More Decks by catla
See All by catla
ベイズ深層学習(6.3)
catla
2
240
ベイズ深層学習(6.2)
catla
3
240
[読み会資料] Federated Learning for Vision-and-Language Grounding Problems
catla
0
340
ベイズ深層学習(5.1~5.2)
catla
0
250
ベイズ深層学習(4.1)
catla
0
480
ベイズ深層学習(3.3~3.4)
catla
19
11k
ベイズ深層学習(2.2~2.4)
catla
6
1.3k
23回アルゴリズムコンテスト 1位解法
catla
6
690
Learning Lightweight Lane Detection CNNs by Self Attention Distillation(ICCV2019)の紹介
catla
0
640
Other Decks in Research
See All in Research
JPA2026_NetworkTutorial_JunKashihara
junkashihara
0
200
OWASP AISVS - C7
shiell
2
790
SoftMatcha 2: 1兆語規模コーパスの超高速かつ柔らかい検索
e869120_sub
7
3.9k
GLIM とMegaParticles:正規分布近似の限界とタイトカップリング&パーティクルフィルタの進展 / GLIM and MegaParticles : Progress of the distribution representation in SLAM
koide3
0
900
Fukui Shibiten 39 - AI Art
butchi
0
200
PGDM: Physically Guided Diffusion Model for L Downscaling
satai
3
500
医療LLMの現在地〜最新研究から社会実装までを考える〜
kento1109
1
1.8k
Anthropic が提案する LLM の内部状態を自然言語で説明可能にした Natural Language Autoencoders / Natural Language Autoencoders Produce Unsupervised Explanations of LLM Activations
shunk031
0
320
敵対生成プロンプト同時探索による内省型プロンプト最適化
kinoue_smarthr
0
430
Vector Map as Language: Toward Unified Remote Sensing Vector Mapping
satai
3
300
量子サマースクール2026「量子計算機アーキテクチャ分野の概観」
youten622
1
940
Karkada さんの論文 × 2 の紹介: (1) Closed-Form Training Dynamics Reveal Learned Features and Linear Structure in Word2Vec-like Models, (2) Symmetry in language statistics shapes the geometry of model representations
eumesy
PRO
1
790
Featured
See All Featured
Balancing Empowerment & Direction
lara
6
1.3k
The agentic SEO stack - context over prompts
schlessera
0
940
AI Search: Where Are We & What Can We Do About It?
aleyda
0
8k
The Language of Interfaces
destraynor
162
27k
Distributed Sagas: A Protocol for Coordinating Microservices
caitiem20
333
23k
How STYLIGHT went responsive
nonsquared
100
6.3k
HU Berlin: Industrial-Strength Natural Language Processing with spaCy and Prodigy
inesmontani
PRO
0
730
The Director’s Chair: Orchestrating AI for Truly Effective Learning
tmiket
1
310
WENDY [Excerpt]
tessaabrams
14
39k
Pawsitive SEO: Lessons from My Dog (and Many Mistakes) on Thriving as a Consultant in the Age of AI
davidcarrasco
0
250
CSS Pre-Processors: Stylus, Less & Sass
bermonpainter
360
31k
Building a Scalable Design System with Sketch
lauravandoore
464
34k
Transcript
5(44BMU*EFOUJpDBUJPO$IBMMFOHF 5FBN-FBSOJOHUIF'VUVSF UIQMBDFTPMVUJPO DBUMB ,BHHMFNFFUVQ LBHHMF@UPLZP
ࣗݾհ ໊લܡɹঘً 5XJUUFS!LBUTVSB@KQ ॴଐஜେֶɹใֶ܈ɹใՊֶྨɹ ϓϩάϥϛϯάྺ ઐ߈ແ͠ ͦͷଞ
ࠓճ͕,BHHMFॳઓ
νʔϜհ
ίϯϖͷ֓ཁ ը૾͔ΒԘؚ͕·Ε͍ͯΔ෦ΛృΓֆ͢Δίϯϖ ηϚϯςΟοΫηάϝϯςʔγϣϯ *OQVUσʔλάϨʔεέʔϧը૾ Y σʔλ EFQUIσʔλը૾ͷਂ͞ใ 5SBJOJOHEBUBTFUຕ 5FTUEBUBTFUຕ ࢀՃνʔϜ
UFBNT σʔλαΠζ͕খ͍͞ ظؒɿd
wධՁࢦඪ w *P6 JOUFSTFDUJPOPWFSVOJPO ίϯϖͷ֓ཁ w͔ͬͨ͠ wॎʹਨͳϚεΫͷ༧ଌਫ਼
1VCMJD-#UI
1SJWBUF-#UI
4PMVUJPO 6/FUϕʔε
4PMVUJPO
4PMVUJPO 6/FU
4PMVUJPO d#BTF&ODPEFSd DBUMB 3FT/F9U MZBLB 3FT/FU BMVQ
3FT/FU 3FT/F9U DPTUBT 3FT/FU PVSBOPT %FOTF/FU USJBO 3FT/FU
4PMVUJPO .BY1PPMJOH $POWE TUSJEFT *OQVUTJ[F Y Y
4PMVUJPOd&ODPEFSd #BDLCPOF#MPDL #BDLCPOF#MPDL Y D D
I X D (MPCBMQPPMJOH %FOTF 3FMV %FOTF 4JHNPJE
4PMVUJPOd&ODPEFSd #BDLCPOF#MPDL #BDLCPOF#MPDL Y Y I X D (MPCBMQPPMJOH %FOTF
D D 3FMV %FOTF 4JHNPJE $POWE LFSOFMTJ[F PVUDIBOOFM 4JHNPJE I X I X D
4PMVUJPOdDFOUFSd $POWE #/ 3FMV $POW % $POWE #/ 3FMV $POWE
#/ 3FMV $POW % $POW % $POW % %%FMBUJPO
4PMVUJPOd%FDPSEFSd $POWE #/ 3FMV $POWE #/ 3FMV Y Y
D D $POWE LFSOFMTJ[F PVUDIBOOFM 4JHNPJE I X I X D $POWE #/ 3FMV $POWE #/ 3FMV I X D (MPCBMQPPMJOH %FOTF 3FMV %FOTF 4JHNPJE
4PMVUJPO -PTT 'PDBM-PTT -PWBT[MPTT BDUJWBUJPO3FMVPSFMV 0QUJNJ[FS "EBN NPNFOUVN4(%
4PMVUJPO )ZQFSDPMNO
4PMVUJPO ɾQTFVEPMBCFMJOH ɾ4OBQ4IPU&OTFNCMJOH ɾ'BTU(FPNFUSJD&OTFNCMJOH
4PMVUJPO
4PMVUJPO -PTT
4PMVUJPOd1PTU1SPDFTTJOHd
4PMVUJPOd1PTU1SPDFTTJOHd
4PMVUJPOd1PTU1SPDFTTJOHd 1VCMJD-# 1SJWBUF-#ʹɺ ӨڹΛ༩͑ͳ͔ͬͨ
͓ΘΓʹ 5XJUUFS աڈͷίϯϖ 2JJUB (JU)VC .FEJVN 4MJEFTIBSF BS9JW ຊ FUDʜ
ͲΜͲΜ׆༻͠Α͏ʂ ᩦཉʹ୳ͤ༗༻ͳใ͕͝Ζ͝Ζམͪͯ ·͢ɻࠓճͷίϯϖͰɺ͜ΕΒΛಛʹ׆ ༻͍ͯ͠·ͨ͠ɻͥͻࢀߟʹ