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
文献紹介: Query and Output: Generating Words by Que...
Search
Sponsored
·
Your Podcast. Everywhere. Effortlessly.
Share. Educate. Inspire. Entertain. You do you. We'll handle the rest.
→
Yumeto Inaoka
July 20, 2018
Research
1
200
文献紹介: Query and Output: Generating Words by Querying Distributed Word Representations for Paraphrase Generation
2018/07/20の文献紹介で発表
Yumeto Inaoka
July 20, 2018
Tweet
Share
More Decks by Yumeto Inaoka
See All by Yumeto Inaoka
文献紹介: Quantity doesn’t buy quality syntax with neural language models
yumeto
1
210
文献紹介: Open Domain Web Keyphrase Extraction Beyond Language Modeling
yumeto
0
270
文献紹介: Self-Supervised_Neural_Machine_Translation
yumeto
0
180
文献紹介: Comparing and Developing Tools to Measure the Readability of Domain-Specific Texts
yumeto
0
190
文献紹介: PAWS: Paraphrase Adversaries from Word Scrambling
yumeto
0
180
文献紹介: Beyond BLEU: Training Neural Machine Translation with Semantic Similarity
yumeto
0
310
文献紹介: EditNTS: An Neural Programmer-Interpreter Model for Sentence Simplification through Explicit Editing
yumeto
0
380
文献紹介: Decomposable Neural Paraphrase Generation
yumeto
0
250
文献紹介: Analyzing the Limitations of Cross-lingual Word Embedding Mappings
yumeto
0
260
Other Decks in Research
See All in Research
教師あり学習と強化学習で作る 最強の数学特化LLM
analokmaus
2
950
ペットのかわいい瞬間を撮影する オートシャッターAIアプリへの スマートラベリングの適用
mssmkmr
0
370
英語教育 “研究” のあり方:学術知とアウトリーチの緊張関係
terasawat
1
510
SkySense V2: A Unified Foundation Model for Multi-modal Remote Sensing
satai
3
630
AI Agentの精度改善に見るML開発との共通点 / commonalities in accuracy improvements in agentic era
shimacos
6
1.4k
When Learned Data Structures Meet Computer Vision
matsui_528
1
4k
FUSE-RSVLM: Feature Fusion Vision-Language Model for Remote Sensing
satai
3
210
AIスーパーコンピュータにおけるLLM学習処理性能の計測と可観測性 / AI Supercomputer LLM Benchmarking and Observability
yuukit
1
730
台湾モデルに学ぶ詐欺広告対策:市民参加の必要性
dd2030
0
200
Ankylosing Spondylitis
ankh2054
0
150
Tiaccoon: Unified Access Control with Multiple Transports in Container Networks
hiroyaonoe
0
1.1k
YOLO26_ Key Architectural Enhancements and Performance Benchmarking for Real-Time Object Detection
satai
3
110
Featured
See All Featured
Utilizing Notion as your number one productivity tool
mfonobong
4
260
SEO in 2025: How to Prepare for the Future of Search
ipullrank
3
3.4k
Balancing Empowerment & Direction
lara
5
940
VelocityConf: Rendering Performance Case Studies
addyosmani
333
24k
Design and Strategy: How to Deal with People Who Don’t "Get" Design
morganepeng
133
19k
A Modern Web Designer's Workflow
chriscoyier
698
190k
Taking LLMs out of the black box: A practical guide to human-in-the-loop distillation
inesmontani
PRO
3
2.1k
Designing for Timeless Needs
cassininazir
0
160
Believing is Seeing
oripsolob
1
81
Mind Mapping
helmedeiros
PRO
1
120
Beyond borders and beyond the search box: How to win the global "messy middle" with AI-driven SEO
davidcarrasco
3
70
Digital Ethics as a Driver of Design Innovation
axbom
PRO
1
210
Transcript
Query and Output: Generating Words by Querying Distributed Word Representations
for Paraphrase Generation Shuming Ma, Xu Sun, Wei Li, Sujian Li, Wenjie Li, Xuancheng Ren. Proceedings of NAACL-HLT 2018, pages 196-206, 2018. จݙհ Ԭٕज़Պֶେֶࣗવݴޠॲཧݚڀࣨ ҴԬເਓ
"CTUSBDU wطଘͷ4FRTFRϞσϧ͕ੜ͢Δจ จ๏తʹਖ਼͍͕͠ҙຯతʹෆదͳ͜ͱ͕Α͋͘Δ w୯ޠࢄදݱͷݕࡧʹΑͬͯ୯ޠΛੜ͢Δ 8PSE&NCFEEJOH"UUFOUJPO/FUXPSL 8&"/ ΛఏҊ wݴ͍͕͑ॏཁͳςΩετฏқԽͱจཁλεΫͰ TUBUFPGUIFBSUΛୡ
!2
*OUSPEVDUJPO w ैདྷͷ4FRTFRϞσϧ୯ޠͷҙຯͰͳ͘܇࿅ ηοτͷ୯ޠύλʔϯΛ҉ه͢Δ͕͋Δ ˡσίʔμͷग़ྗ͕ҙຯతใΛϞσϦϯά ɹ͍ͯ͠ͳ͍ͨΊ w σίʔμͷग़ྗ͕࣋ͭύϥϝʔλ͕ଟ͍ ӅΕͷ࣍ݩ͕ ޠኮαΠζ͕ສͷ߹
ύϥϝʔλ ສͱͳΔ !3
8&"/ w 8PSE&NCFEEJOH"UUFOUJPO/FUXPSL w 3//ͷग़ྗΛΫΤϦͱͯ͠࠷Ұக͢Δࢄදݱ ͷ୯ޠΛBUUFOUJPOΛ༻ͨ͠ݕࡧʹΑͬͯબ w ୯ޠͷࢄදݱΤϯίʔμɺσίʔμͷೖྗ ʹՃ͑ͯग़ྗͷΫΤϦʹΑͬͯߋ৽͞ΕΔ !4
!5 8&"/
!6 8&"/
!7 8&"/
!8 8&"/
ΫΤϦͱ୯ޠͷηοτ͔ΒείΞΛܭࢉ RUλΠϜεςοϓUͷRVFSZ FJJ൪ͷީิ୯ޠ J ʜ O OޠኮαΠζ είΞ͕࠷େͱͳΔ୯ޠΛબ WBMJEBUJPOTFUTͰͷੑೳΛجʹHFOFSBMΛ༻
!9
5SBJOJOH w ୯ޠͷબʹݕࡧΛ༻͍͍ͯΔ͕ɺҰൠతͳ 4FRTFRͱಉ༷ʹඍՄೳ ˠଛࣦؔಉ͡ͷ͕༻Մೳ w "EBN Ћ Ќ Ќ
ЏF !10
&YQFSJNFOUT 5FYU4JNQMJpDBUJPO w %BUBTFUT 1BSBMMFM8JLJQFEJB4JNQMJpDBUJPO$PSQVT 18,1 USBJOWBMJEUFTU
&OHMJTI8JLJQFEJBBOE4JNQMF&OHMJTI8JLJQFEJB &84&8 L USBJOWBMJEUFTU UFTUTFU".5ͰಘΒΕͨͭͷ3FGFSFODFΛ࣋ͭ !11
&YQFSJNFOUT 5FYU4JNQMJpDBUJPO w &WBMVBUJPO.FUSJDT #-&6 ػց༁ฏқԽͰ͘༻͍ΒΕ͍ͯΔࣗಈධՁख๏ ਓखධՁ ྲྀெੑɺଥੑɺฏқ͞ΛͰධՁ ฏқ͞ग़ྗ͕ೖྗͱൺͯͲΕ͚ͩฏқ͔Λࣔ͢
!12
݁Ռ ࣗಈධՁ !13
݁Ռ ਓखධՁ !14
"OBMZTJT w 8&"/ैདྷͷ4FRTFRͱൺͯύϥϝʔλ͕গͳ͍ !15
"OBMZTJT w /54XW1#.53ඞਢͷཁૉΛ͍͍ܽͯΔ w 4#.54"3*ྲྀெ͕ͩҙຯ͕ҟͳΔ !16
"OBMZTJT !17
"OBMZTJT !18 ˢলུ͕ଟ͘ใ͕ෆ͍ͯ͠Δ
"OBMZTJT !19 ˢTJFNFOTNBSUJO SSC TIVSCBͱ͍ͬͨແؔͷ ɹ୯ޠΛग़ྗ
"OBMZTJT !20 ˣྲྀெ͕ͩҙຯ͕ҟͳΓΑΓཧղ͕͘͠ͳ͍ͬͯΔ
$PODMVTJPO w ΫΤϦʹΑΔ୯ޠࢄදݱͷݕࡧ͔Β୯ޠΛੜ ͢ΔFODPEFSEFDPEFSGSBNFXPSLΛఏҊ w ͭͷӳޠฏқԽσʔληοτʹ͓͍ͯ ϕʔεϥΠϯͱൺֱͯ͠#-&6͕ͦΕͧΕ ͓Αͼ্ͨ͠ w ຊϞσϧTUBUFPGUIFBSUΛୡ͍ͯ͠Δ
!21