What are Generative Models? | VAE & GAN | Intro to AI
I'm Zhuoyue, a senior undergrad from the Computer Science department at the University of Toronto, also a software engineer at IBM Watson. This is my first video for the "Intro to AI" series, where I talked about Variational Autoencoder (VAE) and Generative Adversarial Network (GAN).
0:00 Generative Models & Unsupervised Learning
1:02 Variational Autoencoder (VAE) & Autoencoder (AE)
2:42 Generative Adversarial Network (GAN)
3:32 My Thoughts about the Latent Space
This video assumes that you know what Deep Learning is. If you don't, don't worry, here is a great video from MIT: https://youtu.be/5tvmMX8r_OM?t=289. In you are interested in research, here are several cool projects I came across:
1. Ali, Safinah, Daniella DiPaola, and Cynthia Breazeal. "What are GANs?: Introducing Generative Adversarial Networks to Middle School Students." (2021)
This work presents an educational module that teaches middle school students how GANs work and how they can create media using GANs. http://robotic.media.mit.edu/wp-content/uploads/sites/7/2021/03/EAAI-What-are-GANs_.pdf
2. Roberts, Adam, et al. "MusicVAE: Creating a palette for musical scores with machine learning, March 2018." (2018).
This work presents a VAE model for music generation, check out their link for some cool melody interpolation demo https://magenta.tensorflow.org/music-vae
3. Engel, Jesse, et al. "Gansynth: Adversarial neural audio synthesis." arXiv preprint arXiv:1902.08710 (2019).
This work presents a GAN model for timbre interpolation, check out their website for the demo https://magenta.tensorflow.org/gansynth
------------------------------------------------------------------------------------------------------------------------------------------
Producer | Zhuoyue Lyu
Video Shooting | Zhuoyue Lyu
Post Production | Zhuoyue Lyu
Background music is "Clear Skies" credit to Andy Ellison, https://app.soundstripe.com/artists/548. Some contents and pictures are taken from the following sources.
References:
[1] Fei-Fei Li and Justin Johnson and Serena Yeung, Stanford University, Lecture 13 | Generative Models, https://youtu.be/5WoItGTWV54
[2] Jabril Ashe et al., Unsupervised Learning: Crash Course AI #6, https://youtu.be/JnnaDNNb380
[3] Ava Soleimany and Alexander Amini, MIT 6.S191: Deep Generative Modeling, https://youtu.be/BUNl0To1IVw, http://introtodeeplearning.com/slides/6S191_MIT_DeepLearning_L4.pdf
[4] Ava Soleimany and Alexander Amini, Barack Obama: Intro to Deep Learning | MIT 6.S191 https://youtu.be/5tvmMX8r_OM
[5] Pascal Janetzky, Generative Networks: From AE to VAE to GAN to CycleGAN https://laptrinhx.com/generative-networks-from-ae-to-vae-to-gan-to-cyclegan-1119555432/
[6] Irhum Shafkat, Intuitively Understanding Variational Autoencoders https://towardsdatascience.com/intuitively-understanding-variational-autoencoders-1bfe67eb5daf
[7] Google, Overview of GAN Structure, https://developers.google.com/machine-learning/gan
[8] David Duvenaud and Jesse Bettencourt, CSC412 Winter 2020: Probabilistic Machine Learning, https://probmlcourse.github.io/csc412/lectures/week_11/
[9] Throneful, PlayerUnknown's Battlegrounds (PUBG) (2021) - Gameplay (PC UHD) https://youtu.be/lsXMumRLo6I
[10] Y. LeCun, L. Bottou, Y. Bengio, and P. Haffner. "Gradient-based learning applied to document recognition." Proceedings of the IEEE, 86(11):2278-2324, November 1998. http://yann.lecun.com/exdb/mnist/
[11] Richard Gall, Working Principles of Generative Adversarial Networks (GANs), https://dzone.com/articles/working-principles-of-generative-adversarial-netwo
[12] Alexander Mordvintsev, Michael Tyka and Christopher Olah, Deep Dream, https://ai.googleblog.com/2015/06/inceptionism-going-deeper-into-neural.html
I’m Zhuoyue, a senior undergrad from the Computer Science department at the University of Toronto, also a software engineer at IBM Watson. This is my first video for the “Intro to AI” series, where I talked about Variational Autoencoder (VAE) and Generative Adversarial Network (GAN).
0:00 Generative Models & Unsupervised Learning
1:02 Variational Autoencoder (VAE) & Autoencoder (AE)
2:42 Generative Adversarial Network (GAN)
3:32 My Thoughts about the Latent Space
This video assumes that you know what Deep Learning is. If you don’t, don’t worry, here is a great video from MIT: https://youtu.be/5tvmMX8r_OM?t=289. In you are interested in research, here are several cool projects I came across:
1. Ali, Safinah, Daniella DiPaola, and Cynthia Breazeal. “What are GANs?: Introducing Generative Adversarial Networks to Middle School Students.” (2021)
This work presents an educational module that teaches middle school students how GANs work and how they can create media using GANs. http://robotic.media.mit.edu/wp-content/uploads/sites/7/2021/03/EAAI-What-are-GANs_.pdf
2. Roberts, Adam, et al. “MusicVAE: Creating a palette for musical scores with machine learning, March 2018.” (2018).
This work presents a VAE model for music generation, check out their link for some cool melody interpolation demo https://magenta.tensorflow.org/music-vae
3. Engel, Jesse, et al. “Gansynth: Adversarial neural audio synthesis.” arXiv preprint arXiv:1902.08710 (2019).
This work presents a GAN model for timbre interpolation, check out their website for the demo https://magenta.tensorflow.org/gansynth
——————————————————————————————————————————————
Producer | Zhuoyue Lyu
Video Shooting | Zhuoyue Lyu
Post Production | Zhuoyue Lyu
Background music is “Clear Skies” credit to Andy Ellison, https://app.soundstripe.com/artists/548. Some contents and pictures are taken from the following sources.
References:
[1] Fei-Fei Li and Justin Johnson and Serena Yeung, Stanford University, Lecture 13 | Generative Models, https://youtu.be/5WoItGTWV54
[2] Jabril Ashe et al., Unsupervised Learning: Crash Course AI #6, https://youtu.be/JnnaDNNb380
[3] Ava Soleimany and Alexander Amini, MIT 6.S191: Deep Generative Modeling, https://youtu.be/BUNl0To1IVw, http://introtodeeplearning.com/slides/6S191_MIT_DeepLearning_L4.pdf
[4] Ava Soleimany and Alexander Amini, Barack Obama: Intro to Deep Learning | MIT 6.S191 https://youtu.be/5tvmMX8r_OM
[5] Pascal Janetzky, Generative Networks: From AE to VAE to GAN to CycleGAN https://laptrinhx.com/generative-networks-from-ae-to-vae-to-gan-to-cyclegan-1119555432/
[6] Irhum Shafkat, Intuitively Understanding Variational Autoencoders https://towardsdatascience.com/intuitively-understanding-variational-autoencoders-1bfe67eb5daf
[7] Google, Overview of GAN Structure, https://developers.google.com/machine-learning/gan
[8] David Duvenaud and Jesse Bettencourt, CSC412 Winter 2020: Probabilistic Machine Learning, https://probmlcourse.github.io/csc412/lectures/week_11/
[9] Throneful, PlayerUnknown’s Battlegrounds (PUBG) (2021) – Gameplay (PC UHD) https://youtu.be/lsXMumRLo6I
[10] Y. LeCun, L. Bottou, Y. Bengio, and P. Haffner. “Gradient-based learning applied to document recognition.” Proceedings of the IEEE, 86(11):2278-2324, November 1998. http://yann.lecun.com/exdb/mnist/
[11] Richard Gall, Working Principles of Generative Adversarial Networks (GANs), https://dzone.com/articles/working-principles-of-generative-adversarial-netwo
[12] Alexander Mordvintsev, Michael Tyka and Christopher Olah, Deep Dream, https://ai.googleblog.com/2015/06/inceptionism-going-deeper-into-neural.html
