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
MixPoet
Search
Zhang Yixiao
April 30, 2020
Research
440
4
Share
Embed
Copy iframe code
Copy JS code
Copy link
Start on current slide
MixPoet
Zhang Yixiao
April 30, 2020
More Decks by Zhang Yixiao
See All by Zhang Yixiao
CoCon
ldzhangyx
0
390
vq-cpc
ldzhangyx
0
380
diora
ldzhangyx
0
290
drummernet
ldzhangyx
0
250
ON-LSTM
ldzhangyx
0
210
Other Decks in Research
See All in Research
Harness Engineering and Al Agent
kzinmr
3
1.7k
言語モデルから言語について語る際に押さえておきたいこと
eumesy
PRO
5
2.3k
CyberAgent AI Lab研修 / Social Implementation Anti-Patterns in AI Lab
chck
7
4.7k
Any-Optical-Model: A Universal Foundation Model for Optical Remote Sensing
satai
3
840
討議:RACDA設立30周年記念都市交通フォーラム2026
trafficbrain
0
960
NII S. Koyama's Lab Research Overview AY2026
skoyamalab
0
320
通時的な類似度行列に基づく単語の意味変化の分析
rudorudo11
0
320
Dual Quadric表現を用いた動的物体追跡とRGB-D・IMU制約の密結合によるオドメトリ推定
nanoshimarobot
0
410
Unified Audio Source Separation (Defense Slides)
kohei_1979
1
620
Apache Gravitinoで実現する Icebergカタログ統合とアクセスの一元化
matsumooon
0
290
Model Discovery and Graph Simulation: A Lightweight Gateway to Chaos Engineering
anatolykr
0
200
AIエージェント時代のLLM-jpモデルのあるべき姿
k141303
0
480
Featured
See All Featured
Hiding What from Whom? A Critical Review of the History of Programming languages for Music
tomoyanonymous
2
860
The #1 spot is gone: here's how to win anyway
tamaranovitovic
2
1.1k
Why Mistakes Are the Best Teachers: Turning Failure into a Pathway for Growth
auna
0
160
Code Review Best Practice
trishagee
74
20k
More Than Pixels: Becoming A User Experience Designer
marktimemedia
3
440
Design and Strategy: How to Deal with People Who Don’t "Get" Design
morganepeng
133
19k
WENDY [Excerpt]
tessaabrams
11
38k
The Illustrated Guide to Node.js - THAT Conference 2024
reverentgeek
1
390
Jess Joyce - The Pitfalls of Following Frameworks
techseoconnect
PRO
1
170
How To Stay Up To Date on Web Technology
chriscoyier
790
250k
Thoughts on Productivity
jonyablonski
76
5.2k
Ruling the World: When Life Gets Gamed
codingconduct
0
260
Transcript
MixPoet: Diverse Poetry Generation via Learning Controllable Mixed Latent Space
ArXiv: 2003.06094v1 Presenter: Yixiao Zhang
Overview • Idea: 诗人经历、历史背景等 => 诗歌风格多样化 • Methods: • semi-supervised
VAE • disentangling latent space to sub-spaces • each sub-space corresponds to one factor conditioning • adversarial training
Introduction • 近年的研究,主要考虑语义连贯、主题相关 • 存在diversity的困扰 • diversity: • 主题间多样性:给定两个topic words,生成不同的诗歌
• 主题内多样性:给定一个topic word,生成不同的诗歌 • * 现有的模型倾向于记住常见pattern
Introduction • 生活经历、历史背景、文学流派 => 影响风格
Introduction • MixPoet: semi-supervised VAE • 将latent space分解为sub-spaces,与影响因子一一对应 • 训练阶段:模型预测无label诗歌的factors
• 测试阶段:指定factor的值,生成风格化的诗歌
Related Work • 诗歌生成模型 (RNNs, Memory Models, etc. ) •
多样性的先前研究: • MRL system: 强化学习,鼓励选用高TF-IDF的词汇 • USPG: 无监督最大化style vector和诗歌的mutual information
Related Work • VAE文本生成/诗歌生成 • Yang et. al, 2018b: 学习context-conditioned
latent variable • Hu et al. 2017: 对生成的诗歌进行对抗训练,增强topic相关性 • CVAE 对话多样性: Learning Discourse-level Diversity for Neural Dialog Models using Conditional Variational Autoencoders, ACL 2017 • 本文的对抗:在latent space上做对抗训练
Method • topic keyword: mixture empirical distributions: labeled/ unlabeled
Method: Generator • GRU based model • 是length embedding
Method: Semi-supervised C-VAE • 目的是学习 • 引入z • 由于style与semantics耦合 •
不假设y与z的独立性,而是: • 顺序: w => y => z => x (无y label时)
Method: Semi-supervised C-VAE • then for labeled data: • 估计先验
• 和后验 分别使用一个network计算, recon时最小化KL散度。
Method: Semi-supervised C-VAE • labeled data is too limited •
将y看作另一个latent variable • 估计先验 • 和后验 分别使用一个MLP network计算, recon y时最小化KL散度。
Method: Semi-supervised C-VAE • Total Loss:
Method: Latent Space Mixture • 多个factor时的情形: • 独立性假设:
Method: Latent Space Mixture • How to learn mixed latent
space? • For Isotropic Gaussian Space:
Method: Latent Space Mixture • How to learn mixed latent
space? • For Universal Space: 对于condition: ita是噪声,delta是脉冲函数,c是w, y => 从分布中sample出一个值
Method: Latent Space Mixture • 之后使得discriminator区分这两个z • 估计KL散度: • 其中
就是discriminator
Experiments • factors: • 军旅生涯, 乡村生活, 其他 • 时代繁荣, 时代衰落
• => 6种style
Experiments • Baseline: • Ground Truth • C-VAE • USPG
• MRL: SOTA • fBasic, 监督学习模型
Experiments • 多样性,使用Jaccard Similarity指数评价,越低越好 • 诗歌质量:使用Language Model Score(LMS)评价 • 观察:
• 大多数模型倾向生成重复的短语 • MRL与Basic在intra部分只能生成极其相似的诗歌 • C-VAE情况类似
Experiments • Factor Control Results: • 测试生成的诗歌是否与给定因子类别一致
Experiments • 主观实验
Analysis: Style Mixture
Analysis