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
Sponsored
·
Your Podcast. Everywhere. Effortlessly.
Share. Educate. Inspire. Entertain. You do you. We'll handle the rest.
→
Zhang Yixiao
April 30, 2020
Research
4
420
MixPoet
Zhang Yixiao
April 30, 2020
Tweet
Share
More Decks by Zhang Yixiao
See All by Zhang Yixiao
CoCon
ldzhangyx
0
370
vq-cpc
ldzhangyx
0
360
diora
ldzhangyx
0
270
drummernet
ldzhangyx
0
230
ON-LSTM
ldzhangyx
0
180
Other Decks in Research
See All in Research
データサイエンティストの業務変化
datascientistsociety
PRO
0
330
生成AI による論文執筆サポート・ワークショップ データ分析/論文ドラフト編 / Generative AI-Assisted Paper Writing Support Workshop: Data Analysis and Drafting Edition
ks91
PRO
0
110
明日から使える!研究効率化ツール入門
matsui_528
10
5.7k
その推薦システムの評価指標、ユーザーの感覚とズレてるかも
kuri8ive
1
350
20年前に50代だった人たちの今
hysmrk
0
180
SREはサイバネティクスの夢をみるか? / Do SREs Dream of Cybernetics?
yuukit
3
450
Ankylosing Spondylitis
ankh2054
0
150
Any-Optical-Model: A Universal Foundation Model for Optical Remote Sensing
satai
3
270
The Landscape of Agentic Reinforcement Learning for LLMs: A Survey
shunk031
3
440
【NICOGRAPH2025】Photographic Conviviality: ボディペイント・ワークショップによる 同時的かつ共生的な写真体験
toremolo72
0
200
病院向け生成AIプロダクト開発の実践と課題
hagino3000
0
590
「車1割削減、渋滞半減、公共交通2倍」を 熊本から岡山へ@RACDA設立30周年記念都市交通フォーラム2026
trafficbrain
1
840
Featured
See All Featured
Performance Is Good for Brains [We Love Speed 2024]
tammyeverts
12
1.5k
From π to Pie charts
rasagy
0
160
End of SEO as We Know It (SMX Advanced Version)
ipullrank
3
4.1k
Information Architects: The Missing Link in Design Systems
soysaucechin
0
850
SEO Brein meetup: CTRL+C is not how to scale international SEO
lindahogenes
1
2.5k
Let's Do A Bunch of Simple Stuff to Make Websites Faster
chriscoyier
508
140k
個人開発の失敗を避けるイケてる考え方 / tips for indie hackers
panda_program
122
21k
Building Applications with DynamoDB
mza
96
7k
Product Roadmaps are Hard
iamctodd
PRO
55
12k
Cheating the UX When There Is Nothing More to Optimize - PixelPioneers
stephaniewalter
287
14k
How to Grow Your eCommerce with AI & Automation
katarinadahlin
PRO
1
160
16th Malabo Montpellier Forum Presentation
akademiya2063
PRO
0
87
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