Completed
Fundamental difference between ICA and PCA
Class Central Classrooms beta
YouTube videos curated by Class Central.
Classroom Contents
Nonlinear Independent Component Analysis - Aapo Hyvärinen
Automatically move to the next video in the Classroom when playback concludes
- 1 Intro
- 2 Abstract
- 3 Success of Artificial Intelligence
- 4 Neural networks
- 5 Deep learning
- 6 Importance unsupervised learning
- 7 ICA as principled unsupervised learning
- 8 Fundamental difference between ICA and PCA
- 9 Identifiability means ICA does blind source separation
- 10 Example of ICA: Brain source separation
- 11 Example of ICA: Image features
- 12 Nonlinear ICA is an unsolved problem
- 13 Darmois construction
- 14 Temporal structure helps in nonlinear ICA
- 15 Algorithmic trick: "Self-supervised" learning
- 16 Theorem: TCL estimates nonlinear nonstationary ICA Assume data follows nonlinear ICA model (t)-f s(tl) with
- 17 Permutation contrastive learning (Hyvärinen and Morioka 2017)
- 18 Illustration of demixing capability by PCL Non Gaussian AR model for sources
- 19 Extensions of nonlinear ICA on time series
- 20 General framework: Deep Latent Variable Models
- 21 Conditioning makes DLVM identifiable
- 22 Alternative approaches to DLVM case
- 23 Conclusion
