Skip to content

BICF Machine Learning 1 - Nanocourse

Have you heard about machine learning and feel it may be of some value for your research? This nanocourse will introduce machine learning from the ground up. No prior experience in machine learning is necessary.

Topics for this two day (all-day) course will include:

  • Intro to machine learning
  • Supervised learning including:
    • regression
    • classification
  • Regularization to handle overfitting
  • Intro to neural networks
  • Intro to convolutional neural networks for image analysis
  • Unsupervised learning including
    • clustering
    • dimensionality reduction
  • Probabilistic graphical models (PGM)
    • fundamentals of the PGM formalism
    • representation with PGMs
    • learning model parameters
    • inference from the model

The course will be interactive, with lectures followed by hands-on learning and exercises. Familiarity with basic programming/scripting concepts is assumed as is some prior programming experience in python.

You will not need to bring a laptop computer for this course. When you arrive you will loginto a prepared account on the BioHPC.

Course Instructors:

Course Administration: Rebekah Craig

Preparation for Class

Schedule

Day 1 | Feb 28th, 2019
Room NB2.100A

Schedule_Day1

Day 2 | March 1st, 2019
Room NG3.202

Probabilistic graphical models: lecture, exercise

Schedule_Day2

TAs:

Additional Resources

Resources for the general course: - Offical Python Tutorial - Jupyter notebook Cheatsheet

Resources for the "probabilistic graphical models" unit:

Theory: - David Blei's "Foundations of Graphical Models" @ Columbia - Eric Xing's "Probabilistic Graphical Models" @ CMU

Probabilistic programming languages/packages for implementation: - STAN [programming language] - PyMC3 [Theano-based package] - Edward [Tensorflow-based package]