User Tools

Site Tools


start

This is an old revision of the document!


Schedule

Links to MS Teams Events:

  1. Office Hours with Chuck: Wednesdays, 10:00 - 11:00 AM
  2. Office Hours with Dejan: Mondays, 1:00 - 3:00 PM and Wednesdays, 3:00 - 5:00

Lecture videos are available from the Canvas home page.

To use jupyter notebooks on our CS department machines, you must add this line to your .bashrc file:

export PATH=/usr/local/anaconda/bin:$PATH

This is a tentative schedule of CS545 topics for Fall, 2020. This will be updated during the summer and as the fall semester continues.

August

Week Topic Material Reading Assignments
Week 1:
Aug 24 - Aug 28
Overview of course and the machine learning field. Reminder of how python is used in machine learning. 01 Introduction to CS545
02 Searching for Good Weights in a Linear Model
From Python to Numpy, Chapters 1 - 2
Scipy Lectures, Section 1
Visualization with Matplotlib
Deep Learning, Chapters 1 - 5.1.4

September

Week Topic Material Reading Assignments
Week 2:
Aug 31 - Sept 4
Help with A1. Review of gradients. Gradient descent with SGD, Adam and SCG, 03 Fitting Simple Models Using Gradient Descent in the Squared Error A1.4 Polynomial Model due Friday, Sept 4th, at 10:00 PM
Week 3:
Sept 7 - Sept 11
Implementing neural networks with numpy to predict real-valued variables. Deriving gradients. 04 Scaled Conjugate Gradient
05 Introduction to Gradient Descent for Neural Networks
Week 4:
Sept 14 - Sept 18
Error gradients for neural networks as matrix equations. Discussion of A2.
Introduction to dashboards with python using streamlit.
06 Introduction to Streamlit streamlit.io
Week 5:
Sept 21 - 25
Use of Optimizers for neural networks. Introduction to Pytorch and automatic differentation. 07 Collect Weights in Vector for Optimizers
08 Pytorch autograd, nn.Module
A2.2 Multilayer Neural Network due Friday, Sept 25th, at 10:00 PM. Good examples of solutions are available here.

October

Week Topic Material Reading Assignments
Week 6:
Sept 28 - Oct 2
Neural Network class. 09 Initial Steps towards Defining a NeuralNetwork Class
Week 7:
Oct 5 - Oct 9
Oct 8 Lecture will not meet, but recording will be available.
Help with A3. Dimensionality reduction. 10 Help with A3
11 Low-Dimensional Representations of Data
A3.3 Neural Network Class due Monday, Oct 12, 10:00 PM
Examples of good solutions are available here.
Week 8:
Oct 12 - Oct 16
Brief overview of notes 11.
Introduction to Classification
12 Classification with Neural Networks
Week 9:
Oct 19 - Oct 23
Convolutional neural networks in numpy. 13 NeuralNetwork_Pytorch
14 Introduction to Convolution
Week 10:
Oct 26 - Oct 30
Fully-connected and Convolutional Neural Nets in Pytorch 15 Convolutional Neural Networks
16.1 Convolutional Neural Networks in Pytorch
A4.1 Neural Network Classifier due Tuesday Oct 27, at 10:00 PM
Good examples of solutions are available here.

November

Week Topic Material Reading Assignments
Week 11:
Nov 2 - Nov 6
Comparing network performance. Introduction to Reinforcement Learning. Deep Reinforcement Learning 17 Partitioning Data to Compare Neural Network Performance
18 Introduction to Reinforcement Learning
19 Reinforcement Learning with Neural Network as Q Function
Reinforcement Learning: An Introduction, by Richard Sutton and Andrew Barto, 2nd edition
Week 12:
November 9 - 13
Deep reinforcement learning on simulated physical control problem. 20 Reinforcement Learning to Control a Marble
Week 13:
Nov 16 - Nov 20
A5 Neural Networks in Pytorch due Wednesday, Nov 18 at 10:00 PM
Good examples of solutions are available here.
Nov 23 - Nov 27 Fall Recess!

December

Week Topic Material Reading Assignments
Week 14:
Nov 30 - Dec 4
Clustering.
Support Vector Machines.
22 K-Means Clustering, K-Nearest-Neighbor Classification
23 Support Vector Machines
Week 15:
Dec 7 - Dec 11
Finals Week:
Dec 14 - Dec 18
A6 Reinforcement Learning to Control a Robot due Tuesday, Dec 15th, 10:00 PM. Here is A3mysolution.tar, a neural network implementation you may choose to use for A6.
start.1607017105.txt.gz · Last modified: 2020/12/03 10:38 by 127.0.0.1