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
4
410
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
SREはサイバネティクスの夢をみるか? / Do SREs Dream of Cybernetics?
yuukit
3
420
Earth AI: Unlocking Geospatial Insights with Foundation Models and Cross-Modal Reasoning
satai
3
550
R&Dチームを起ち上げる
shibuiwilliam
1
190
製造業主導型経済からサービス経済化における中間層形成メカニズムのパラダイムシフト
yamotty
0
500
「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
680
病院向け生成AIプロダクト開発の実践と課題
hagino3000
0
560
姫路市 -都市OSの「再実装」-
hopin
0
1.7k
COFFEE-Japan PROJECT Impact Report(海ノ向こうコーヒー)
ontheslope
0
770
"主観で終わらせない"定性データ活用 ― プロダクトディスカバリーを加速させるインサイトマネジメント / Utilizing qualitative data that "doesn't end with subjectivity" - Insight management that accelerates product discovery
kaminashi
16
22k
2026 東京科学大 情報通信系 研究室紹介 (すずかけ台)
icttitech
0
330
[SITA2025 Workshop] 空中計算による高速・低遅延な分散回帰分析
k_sato
0
110
Agentic AI フレームワーク戦略白書 (2025年度版)
mickey_kubo
1
130
Featured
See All Featured
Building Adaptive Systems
keathley
44
2.9k
Ten Tips & Tricks for a 🌱 transition
stuffmc
0
82
RailsConf & Balkan Ruby 2019: The Past, Present, and Future of Rails at GitHub
eileencodes
141
35k
First, design no harm
axbom
PRO
2
1.1k
Exploring anti-patterns in Rails
aemeredith
2
280
Paper Plane
katiecoart
PRO
0
47k
Bioeconomy Workshop: Dr. Julius Ecuru, Opportunities for a Bioeconomy in West Africa
akademiya2063
PRO
1
67
From π to Pie charts
rasagy
0
140
The Anti-SEO Checklist Checklist. Pubcon Cyber Week
ryanjones
0
83
How to build an LLM SEO readiness audit: a practical framework
nmsamuel
1
660
Docker and Python
trallard
47
3.7k
Let's Do A Bunch of Simple Stuff to Make Websites Faster
chriscoyier
508
140k
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