Neural Foundations of Machine Learning

Neural Foundations of Machine Learning
Fall 2025 (GT Metz)

Topics

Projects

Course Material Overview

The class is divided into two exams, and five projects.

	Topic	Exam	Proj	On-Line Lectures	Reading
Week 1 Aug 18	Course Overview History of NN Digital ODE solutions		Proj1 Sept 1	ODE basics Matrix ODEs	VanEssen(1991) VanEssen(2001)
Week 2 Aug 25	Introduction to Neuroscience and Circuits			Neuron Membrane Bio & FET Channel FET Bio Channel Linear Elements Circuit Anal 1^st Circuits Dynamics HighPass 1^st Order	Cortex1
Week 3 Sept 1	Neuroscience, ODEs, Hopfield Networks		Proj2 Sept 22	Nonlinear f( ) NN -- Neurons NN & Current Hopfield I Energy Eff and Phys Compute Intro to Phys Compute	1^st Paper (1982) 1^st TSP (1985) Hopfield + MaxCut Ising H()
Week 4 Sept 8	Exam Prep	Sept 10
Week 5 Sept 15	Unsupervised learning			Unsupervised I Oja's Rule	Oja Alg (1982) ICA: Bell (1995) Sanger1 (1989) Linsker1 (1992)
Week 6 Sept 22	Unsuper learning & Receptive Fields		Proj3 Oct 13		SOM (2013)
Week 7 Sept 29	Winner-Take All, Attention				Lazzaro 1988 Visual Att (1998) Att (Koch) (1985) Att (Itti) (2001)
Week 8 Oct 6	Adaptive Filters Perceptrons, WTA, VQ			Adapt Filt I Adapt Filt II Adapt Filt III	WTA 1 (2000)
Week 9 Oct 13	Supervised Learning Neural structures Backpropagation		Proj4 Nov 3		NN Work (1987) NN Gener (1989)
Week 10 Oct 20	Exam Prep	Oct 22
Break Week	Oct 27-31 .
Week 11 Nov 3	Neuroscience & Larger Networks		Proj5
Week 12 Nov 10	Conv NN & Deep NN
Week 13 Nov 17	Neuro Feedback Generative NN				GAN (2014)
Week 14 Nov 24	ML Physical Implementations			Super-Turing

Useful Figures

Brain regions from paper.

Brain regions from website

Van Essen plot from the original paper.

Cortical Columns from Cortex1

Chapter on MOSFETs from Mead Chp 3, and previous chapter on device Physics from Mead Chp 2.

Projects are to be submitted electronically by 11:59pm Metz time

Last projects is due on Dec 8, submitted electronically by 11:59pm Metz time

Useful Items:

One simulator option is using NENGO that is a Python-based ODE simulation engine (open-source). There are some particular items to help with Neural modeling as well as some useful scripting in Python using these approaches (e.g. paper).
Datasets: A set of MATLAB datasets for Deep Learning. A link in the table to the main page gives the copy command to run the example file.
What is a neural network? Chapter 1, Deep learning https://youtu.be/aircAruvnKk

NN Playground from Google. A useful visualization of a simple NN structure.

The Six Layers of the Cerebral Cortex: Video

Additional circuit related material