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: Decomposable Neural Paraphrase Generation
Search
Sponsored
·
Your Podcast. Everywhere. Effortlessly.
Share. Educate. Inspire. Entertain. You do you. We'll handle the rest.
→
Yumeto Inaoka
July 23, 2019
Research
250
0
Share
文献紹介: Decomposable Neural Paraphrase Generation
2019/07/23の文献紹介で発表
Yumeto Inaoka
July 23, 2019
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
280
文献紹介: Self-Supervised_Neural_Machine_Translation
yumeto
0
190
文献紹介: Comparing and Developing Tools to Measure the Readability of Domain-Specific Texts
yumeto
0
210
文献紹介: PAWS: Paraphrase Adversaries from Word Scrambling
yumeto
0
200
文献紹介: Beyond BLEU: Training Neural Machine Translation with Semantic Similarity
yumeto
0
320
文献紹介: EditNTS: An Neural Programmer-Interpreter Model for Sentence Simplification through Explicit Editing
yumeto
0
410
文献紹介: Analyzing the Limitations of Cross-lingual Word Embedding Mappings
yumeto
0
270
文献紹介: Similarity-Based Reconstruction Loss for Meaning Representation
yumeto
1
240
Other Decks in Research
See All in Research
Scalable dynamic origin-destination demand estimation enhanced by high-resolution satellite imagery data
satai
3
250
「なんとなく」の顧客理解から脱却する ──顧客の解像度を武器にするインサイトマネジメント
tajima_kaho
10
7.6k
COFFEE-Japan PROJECT Impact Report(海ノ向こうコーヒー)
ontheslope
0
1.8k
多様なデータを許容し学習し続ける模倣学習 / Advanced Imitation Learning for VLA
prinlab
0
210
The mathematics of transformers
gpeyre
0
300
LLMアプリケーションの透明性について
fufufukakaka
0
230
明日から使える!研究効率化ツール入門
matsui_528
13
7.2k
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
980
2026.01ウェビナー資料
elith
0
380
第12回人と環境にやさしい交通をめざす全国大会/熊本都市圏「車1割削減、渋滞半減、公共交通2倍」をめざして
trafficbrain
0
110
2026年度 生成AI を活用した論文執筆ガイド/ワークショップ / 2026 Academic Year Guide to Writing Papers Using Generative AI - Workshop
ks91
PRO
0
170
Cross-Media Information Spaces and Architectures
signer
PRO
0
300
Featured
See All Featured
The #1 spot is gone: here's how to win anyway
tamaranovitovic
2
1.1k
Designing for humans not robots
tammielis
254
26k
SEO Brein meetup: CTRL+C is not how to scale international SEO
lindahogenes
1
2.7k
Leading Effective Engineering Teams in the AI Era
addyosmani
9
2k
Abbi's Birthday
coloredviolet
2
7.9k
The Cost Of JavaScript in 2023
addyosmani
55
10k
Distributed Sagas: A Protocol for Coordinating Microservices
caitiem20
333
22k
Navigating the Design Leadership Dip - Product Design Week Design Leaders+ Conference 2024
apolaine
1
340
The Art of Programming - Codeland 2020
erikaheidi
57
14k
A brief & incomplete history of UX Design for the World Wide Web: 1989–2019
jct
2
390
Between Models and Reality
mayunak
4
320
Data-driven link building: lessons from a $708K investment (BrightonSEO talk)
szymonslowik
1
1.1k
Transcript
Decomposable Neural Paraphrase Generation
https://arxiv.org/abs/1906.09741
• • • •
• • •
•
• • •
• • •
• •
• • = [1 , … , ] • =
[1 , … , ]
• • • ℎ = BiLSTM( ; ℎ−1 , ℎ+1
) • = LSTM ℎ , −1 ; −1 • = GumbelSoftmax( , )
• • = − encoderz (, ) • 1:−1 ,
= − encoderz , 1:−1
• • 1:−1 , = σ 1:−1 , ( |1:−1
, )
• 0 , 1 • = LSTM 0 ; 1
; −1 ; −1 • 1:−1 , = GumbelSoftmax ,
• • ∗ = 0 ∗ = 1
• • ℒ = σ=1 log 1:−1 , + σ=1
log ∗ + σ=1 log ( ∗ 1:−1 ,
• • •
• •
•
• •
• •
• •
• •
• •
• • From 1(best) to 4(worst)
• • • • •