SNU Biointelligence Lab

4190.676 Artificial Neural Networks (Fall 2022)

(Artificial Neural Networks, Computational Neuroscience, Computational Models of Intelligence)

Course Objectives

We study “self-learning” networks, i.e. models that learn in an unsupervised and “self-supervised” way without the help of an explicit teacher. These models are neuro-biologically inspired and, usually, self-organizing, dynamic, recurrent, and auto-encoding networks. We examine the principles of neural learning algorithms from the historical models, such as Willshaw-von der Malsburg feature maps, Linsker models, Kohonen’s self-organizing maps, Grossberg models, recurrent networks, Anderson’s brain-state-in-a-box, actor-critic networks, Hopfield’s associative memory, Boltzmann machines, and deep belief networks.

We study mathematical tools for approximation and optimization of the neural learning models. These include information-theoretic algorithms, such as maximum entropy, mutual information, and KL divergence as well as the statistical-mechanical methods, such as Markov chains, Metropolis algorithms, Gibbs sampling, and simulated annealing. We also examine the neurodynamic models of self-supervised, end-to-end learning to solve the challenging problems, such as time series prediction and reconstruction. These include Markov decision processes, approximate dynamic programming, reinforcement learning, sequential Bayesian estimation, Kalman filtering, particle filtering, real-time recurrent learning, dynamic reconstruction of a chaotic process.

Textbooks

Evaluation

Component Weight
Two Exams (Midterm + Final) 80%
Homework 10%
Participation and Discussion 10%

Lecture Schedule

Week Dates Topics Slides
Week 1 9/1 Learning in Neurodynamic Self-organizing Systems — Neural Networks, Unsupervised / Self-supervised Learning; Mathematics for Neural Learning PDF
Week 2 9/6, 9/8 Principal-Components Analysis (Ch. 8) — Principal Component Analysis; Hebbian-Based Maximum Eigenfilter; Hebbian-Based PCA (Ch. 8) — Generalized Hebbian Algorithm; Kernel PCA PDF
Week 3 9/13, 9/15 Self-organizing Maps (Ch. 9) — Willshaw-von der Malsburg Model; Kohonen’s SOM Model PDF
Week 4 9/20, 9/22 Information-Theoretic Learning Models (Ch. 10) — Maximum Entropy, Kullback-Leibler Divergence; Mutual Information (MI) PDF
Week 5 9/27, 9/29 Information-Theoretic Learning Models (Ch. 10) — Infomax, Imax, Imin; Independent Component Analysis (ICA); Statistical-Mechanical Learning Methods (Ch. 11) — Statistical Mechanics, Markov Chains; Metropolis, Gibbs Sampling, Simulated Annealing PDF
Week 6 10/4, 10/6 Deep Neural Networks (Ch. 11) — Boltzmann Machines; Deep Belief Networks PDF
Week 7 10/11, 10/13 Assignment 1 Solving Class; Deep Neural Networks (Ch. 11) — Boltzmann Machines; Deep Belief Networks PDF
Week 8 10/18, 10/20 Summary (10/18); Mid-term Exam (10/20) —
Week 9 10/25, 10/27 No Lecture PDF
Week 10 11/1, 11/3 Dynamic Programming (Ch. 12) — Markov Decision Process, DP, Bellman Equation; ADP, Reinforcement Learning, TD, Q PDF
Week 11 11/8, 11/10 Neurodynamic Models (Ch. 13) — Dynamic Systems, Attractors, Chaos; Hopfield Models, Dynamic Reconstruction PDF
Week 12 11/15, 11/17 Bayesian Filtering (Ch. 14) — State Space Models; Kalman Filters, EKF, CKF PDF
Week 13 11/22, 11/24 Particle Filters (Ch. 14) — Approximate Bayesian Filtering; Particle Filters, SIR Algorithm; Dynamic Recurrent Networks (Ch. 15) — Recurrent Network Architectures; Backpropagation through Time —
Week 14 11/29, 12/1 Real-Time Recurrent Learning (Ch. 15) — RTRL Algorithm, Vanishing Gradients; EKF Algorithm for Training RMLP —
Week 15 12/6, 12/8 Final Exam (12/8) —
Week 16 12/13 Review and Discussion —