Completed
Dynamical systems as a central layer of convergence
Class Central Classrooms beta
YouTube videos curated by Class Central.
Classroom Contents
Deep Learning of Dynamical Systems for Mechanistic Insight and Prediction in Psychiatry
Automatically move to the next video in the Classroom when playback concludes
- 1 Intro
- 2 What is a dynamical system?
- 3 Attractor states in state space
- 4 Working memory tasks & persistent activity
- 5 Memory patterns as attractor states
- 6 Limit cycles ...
- 7 Limit cycles in motor behavior
- 8 Action/ thought sequences as "heteroclinic channels"
- 9 Altered dynamics in psychiatric states
- 10 Dynamical systems as a central layer of convergence
- 11 Recurrent Neural Network → time series
- 12 Making RNN deep in time
- 13 Piecewise-Linear (PL) RNN
- 14 Line-attractor regularization
- 15 Performance on ML benchmarks
- 16 Line-attractors and solving long-range tasks
- 17 Sequential MNIST benchmark
- 18 Generative PLRNN for dynamical systems
- 19 Reconstructing dynamical systems
- 20 Statistical inference for small data: Expectation.Maximization
- 21 Expectation-Maximization Algorithm
- 22 Statistical inference for big data: Sequential VAE & SGVB
- 23 Simple ahead prediction errors may be meaningless
- 24 Reconstructing DS benchmarks
- 25 Reconstructing DS: Lorenz system
- 26 Enforcing line attractor directions helps to capture multiple time scales
- 27 Inferring PLRNN from fMRI data
- 28 Does PLANN really capture measured dynamics?
- 29 Example 1. Unstable neuronal representations in schizophrenia
- 30 Example 2: Inference of dynamical systems from mobile data
- 31 Prediction of medical intervention effects
- 32 Take home's
