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
Integrating Transformer and Paraphrase Rules fo...
Search
onizuka laboratory
December 18, 2018
Research
0
59
Integrating Transformer and Paraphrase Rules for Sentence Simplification
弊研究室で行なったEMNLP2018読み会の発表資料です。
onizuka laboratory
December 18, 2018
Tweet
Share
More Decks by onizuka laboratory
See All by onizuka laboratory
Phrase-Based & Neural Unsupervised Machine Translation
onilab
0
110
Tell-and-Answer: Towards Explainable Visual Question Answering using Attributes and Captions
onilab
0
71
Card-660: A Reliable Evaluation Framework for Rare Word Representation Models
onilab
0
33
A Word-Complexity Lexicon and A Neural Readability Ranking Model for Lexical Simplification
onilab
0
120
An Auto-Encoder Matching Model for Learning Utterance-Level Semantic Dependency in Dialogue Generation
onilab
0
55
Generating More Interesting Responses in Neural Conversation Models with Distributional Constraints
onilab
0
100
Modeling Multi-turn Conversation with Deep Utterance Aggregation
onilab
0
95
Learning Semantic Sentence Embeddings using Pair-wise Discriminator
onilab
0
120
SGM: Sequence Generation Model for Multi-Label Classification
onilab
0
77
Other Decks in Research
See All in Research
Creation and environmental applications of 15-year daily inundation and vegetation maps for Siberia by integrating satellite and meteorological datasets
satai
3
120
Vision And Languageモデルにおける異なるドメインでの継続事前学習が性能に与える影響の検証 / YANS2024
sansan_randd
1
110
定性データ、どう活かす? 〜定性データのための分析基盤、はじめました〜 / How to utilize qualitative data? ~We have launched an analysis platform for qualitative data~
kaminashi
6
1k
When Submarine Cables Go Dark: Examining the Web Services Resilience Amid Global Internet Disruptions
irvin
0
210
心理言語学の視点から再考する言語モデルの学習過程
chemical_tree
2
370
Cross-Media Information Spaces and Architectures
signer
PRO
0
220
Towards a More Efficient Reasoning LLM: AIMO2 Solution Summary and Introduction to Fast-Math Models
analokmaus
2
230
NLP Colloquium
junokim
1
160
クラウドのテレメトリーシステム研究動向2025年
yuukit
3
960
EOGS: Gaussian Splatting for Efficient Satellite Image Photogrammetry
satai
4
270
さくらインターネット研究所 アップデート2025年
matsumoto_r
PRO
0
650
「エージェントって何?」から「実際の開発現場で役立つ考え方やベストプラクティス」まで
mickey_kubo
0
120
Featured
See All Featured
Building an army of robots
kneath
306
45k
The Psychology of Web Performance [Beyond Tellerrand 2023]
tammyeverts
48
2.9k
The MySQL Ecosystem @ GitHub 2015
samlambert
251
13k
The Power of CSS Pseudo Elements
geoffreycrofte
77
5.8k
Optimising Largest Contentful Paint
csswizardry
37
3.3k
What's in a price? How to price your products and services
michaelherold
246
12k
Visualizing Your Data: Incorporating Mongo into Loggly Infrastructure
mongodb
46
9.6k
Six Lessons from altMBA
skipperchong
28
3.9k
We Have a Design System, Now What?
morganepeng
53
7.7k
The Success of Rails: Ensuring Growth for the Next 100 Years
eileencodes
45
7.5k
Building Applications with DynamoDB
mza
95
6.5k
[RailsConf 2023 Opening Keynote] The Magic of Rails
eileencodes
29
9.5k
Transcript
% 4BORJBOH ;IBP 3VJ .FOH %BRJOH)F 4BQUPOP "OEJ 1BSNBOUP
#BNCBOH 5SBOTGPSNFSBOE1BSBQISBTF 3VMFTGPS4FOUFODF4JNQMJGJDBUJPO #ݪ େو &./-1ಡΈձʢʣ
֓ཁ l 5SBOTGPSNFSϕʔεͷΞʔΩςΫνϟ l ฏқԽݴ͍͑σʔλϕʔε 4JNQMF11%#Λ ౷߹͢Δͭͷख๏ʢ%."44ͱ %$44ʣΛఏҊ l ౷߹ͷϝϦοτͭ
◦ จฏқԽͷ405"ΑΓ༏ΕΔ ◦ ϞσϧΑΓਖ਼֬ͳฏқԽنଇΛબ͠Α͏ͱ͢Δ l ιʔείʔυެ։ ◦ IUUQTHJUIVCDPN4BORJBOHUFYU@TJNQMJGJDBUJPO
5SBOTGPSNFS l ࠨ͕Τϯίʔμ l ӈ͕σίʔμ l ଛࣦؔ "#$ = −
log , ɺݱࡏͷϞσϧͷ શͯͷύϥϝʔλ
4JNQMF11%#ͷ౷߹ l ฏқԽͷχϡʔϥϧωοτϫʔΫϞσϧͰɺ ग़ݱස͕ߴ͍ฏқԽنଇΛ܇࿅͢Δ ◦ සنଇΛϊΠζͱଊ͑ͯ͠·͏ l /.5Ͱͷॳͷ֎෦ࣝ౷߹ ◦ ֎෦ࣝΛ౷߹͢Δ4.5ฏқԽྑ͍ʢ9VFUBM
ʣ l 4JNQMF11%# ◦ 1BWMJDL FUBM ◦ ສͷنଇ
ఏҊख๏̍ɿ%$44 l %FFQ$SJUJD4FOUFODF4JNQMJGJDBUJPO.PEFM l ଛࣦؔΛमਖ਼ ◦ සنଇͷݟམͱ͠Λආ͚ΔͨΊʹɺ୯ޠͷੜ֬Λ ฏқԽ֬Ͱ࠶ॏΈ͚ l ྫ
◦ ೖྗɿUIFSFDJQJFOU PGUIFLBUF HSFFOBXBZ NFEBM ◦ ग़ྗɿUIFXJOOFS PGUIFLBUF HSFFOBXBZ NFEBM l SFDJQJFOUΛग़͠ʹ͘͘ɺ XJOOFSΛग़͘͢͠ ͳΔΑ͏ଛࣦΛฦ͍ͨ͠
%$44ͷଛࣦؔ l 012# ฏқԽنଇͷॏΈɺ Ϟσϧύϥϝʔλɺ ೖྗจ l 304543 ಛఆͷ୯ޠͷΈʹɺ"#$ =
− log , ޠኮશମʹযΛ͍ͯͯΔ ◦ ͜ΕΒͷଛࣦؔΛަޓʹ࠷খԽ͢ΔΑ͏܇࿅
ఏҊख๏̎ɿ%."44 l %FFQ.FNPSZ"VHNFOUFE4FOUFODF 4JNQMJGJDBUJPO.PEFM l %$44සنଇ͕ແࢹ͞Ε͍͢ʢ͔͠͠ ܇࿅σʔλ͕ݶΒΕΔ߹ɺසنଇॏཁʣ l ֤نଇʹෳͷΩʔόϦϡʔ༻ϝϞϦΛ༻ҙ ◦
ΩʔϕΫτϧɺίϯςΩετϕΫτϧʢΤϯίʔμͷӅΕ ঢ়ଶͱͦͷ࣌ࠁͷσίʔμӅΕঢ়ଶͷՃॏฏۉʣ ◦ όϦϡʔϕΫτϧɺग़ྗϕΫτϧ
%."44
σʔληοτ l ܇࿅ 8JLJ-BSHF ;IBOHBOE-BQBUB ◦ จର l
ݕূͱධՁɺ5VSL 9VFUBM ◦ ೖྗจʹରͯ͠ਓखͷ̔ϦϑΝϨϯεʢˠྑ࣭ʣ ◦ ݕূ༻ จରɺධՁ༻ จର l ධՁ༻ʹ /FXTFMB ͏ ◦ จର
ධՁࢦඪ l ',(-ʢจͷฏқੑʣ ◦ จͷ͞ͱޠ͔Βܭࢉ l 4"3*ʢՃɾআɾอ࣋͞Ε͍ͯΔ͔ʣ ◦ ೖྗɺग़ྗɺϦϑΝϨϯεΛൺֱ l
نଇར༻ੑʢޠኮมͷਖ਼֬ੑʣ ◦ 4"3*ʹՃͱআΛผʑʹධՁ͢Δ͕ɺ߹Θ͍ͤͨ ◦ ೖྗͱϦϑΝϨϯεΛൺֱ͠ɺฏқԽنଇΛௐͯ ͦΕʹର͢Δ QSFDJTJPOɺSFDBMMɺ'Λܭࢉ
5SBOTGPSNFSͷ݁Ռʢ5VSLʣ l ',(-ͱ 4"3*͕ 3//-45.ΑΓ༏Ε͍ͯͨ l -)͕૿Ճ͢ΔͱείΞ্͕ͬͨ
5SBOTGPSNFSͷ݁Ռʢ5VSLʣ l -)͕૿Ճ͢Δͱ ',(-ͷԼ͕Δ l 4"3*ͷอ͚࣋ͩݮগɺաʹฏқԽͯ͠͠·͏
11%#౷߹ͷ݁Ռʢ5VSLʣ
11%#౷߹ͷ݁Ռʢ5VSLʣ
11%#౷߹ͷ݁Ռʢ5VSLʣ
11%#౷߹ͷ݁Ռʢ/FXTFMBʣ