ML Lecture 18: Unsupervised Learning - Deep Generative Model (Part II)

Показать описание

Hung-yi Lee

Рекомендации по теме

Комментарии

只是我有一个小问题：在将 maximizing likelihood 的时候不是要 minimize minimize 了 KL(q(z|x)||P(x)). lb 上升但是 log(P(x)) 没有上升的情况呢？还是说这两个 KL 的均衡可以让 VAE 在某种程度上获得更好的生成能力呢？

ZijiaTang

老師 37:19右下方mu(x), sigma(x)是否有筆誤是不是該是z?

joanneao

哇塞老師的課真的很前沿而且又很有趣
我們學校的NN相關課程...只能QQ
最近看到Wasserstein GAN熱騰騰
但是數學實在讓我有點崩潰囧

TpBrass

我想選gaussian的原因是N(\mu, \sigma)這個family是一個basis，如果你選別的函數如F(a, b, ... k)，那所有的參數a, b, ... k也要形成basis才能正常運作

rnoro

如果"VAE may just memorize the existing images/linear combination of existing images, instead of generating new images" 那么stable diffusion是如何在VAE的embedding space里面生成全新的图片的呢如果图片足够多embedding dimension足够小那么是不是VAE还是要学习一些东西的呢？

Bhllllll

直观上 VAE 所产生的图, 是一张介于满月与玄月之间的月象图。这是否意味着对输入左下“满月图”来说，用 VAE Encoder（Latent Variable Model）方法成功地对 ”金融领域的时间序列数据“进行预测。谢谢！

jxm

在 30:20 中先調整q(z|x) 讓Lb 等於 algorithm 呢?

jinkuan_lin

还有一个问题： exp(sigma) - (1+sigma) 是在sigma = 0 = 0就好了？为什么还要有这么个loss function呢？

ethanhuang

李老師您好，請問有關 "VAE may just memorize the existing images, instead of generating new images" 這段話的paper 文獻嗎? 我想深入研究，另外 CVAE 在 NLP 上也會有類似的問題嗎?

zeyhvyr

老師請問 6:57 投影片中的m1 m2 m3應該理解成當batch size=3的情況嗎

thenindeed

老师，VAE 和 De-noising auto-encoder 有什么不一样吗？

zhyang

老师的课程很有意思。我有个问题，在VAE第28页PPT里logP（x）替换成log(P(z, x)/P(z|x))是不是弄错了

jacoblee

ML Lecture 18: Unsupervised Learning - Deep Generative Model (Part II)

ML Lecture 18: Unsupervised Learning - Deep Generative Model (Part II)

ML Lecture 17: Unsupervised Learning - Deep Generative Model (Part I)

Unsupervised Learning | Unsupervised Learning Algorithms | Machine Learning Tutorial | Simplilearn

ML Lecture 16: Unsupervised Learning - Auto-encoder

Stanford CS229M - Lecture 18: Unsupervised learning, mixture of Gaussians, moment methods

Lecture 18 | Machine learning

What are Autoencoders?

🔥 Top Machine Learning Projects #Shorts #Simplilearn

How does GPT-3 really work?

(FS 2021) Lecture 14: Unsupervised Learning for Medical Image Analysis

CMU Neural Nets for NLP 2019 (18): Unsupervised and Semi-supervised Learning of Structure

But what is a neural network? | Chapter 1, Deep learning

ML Lecture 15: Unsupervised Learning - Neighbor Embedding

18. Unsupervised and Descriptive Learning -- machine learning

Unsupervised Learning

Deep Learning for Newbies: Supervised or Unsupervised? 🛤️🧠 - Topic 013 #ai #ml

Local Aggregation for Unsupervised Learning of Visual Embeddings

Lecture 16: K-Means (Unsupervised Learning)

MIT CompBio Lecture 12 - Deep Learning

#109 What is PCA| Data Science | Machine Learning Interview Question | Shorts

ML Lecture 14: Unsupervised Learning - Word Embedding

Piotr Bojanowski: Unsupervised methods for deep learning and applications

Computer Scientist Explains Machine Learning in 5 Levels of Difficulty | WIRED

Supervised, Unsupervised and Reinforcement Learning in Artificial Intelligence in Hindi