Week07

Feedforward Neural Networks and Back-Propagation

Interactive session, Monday March 10

• Screencast part 1
• Screencast part 2

Weekly lecture:

Slides: feedforward neural networks

Lecture videos:

1. Feed-forward neural networks (Multi-layer Perceptrons)
2. Matrix representations
3. The Back-propagation Algorithm
   1. The basics
   2. Linear Regression
   3. The first layer
   4. Logistic Regression
   5. Multi-class classification
4. Finer Details
5. Evaluation: Precision, Recall and F1

Syllabus:

• Marsland, Machine Learning
  • Chapter 4 "The Multi-layer Perceptron", everything, except
    • (for now) 4.2.6 Picking up momentum
    • the technical details of 4.6 Deriving Back-propagation,
      What you have to know here, is covered by the lectures.
• Speech and Language Processing. Daniel Jurafsky & James H. Martin.
  • Chapter 7, sections 7.1, 7.2, 7.3, 7.4, 7.5
• If you feel that your skills with vectors, matrices and tensors need some refreshment, check out our linear algebra cheat-sheet or tutorials.

Weekly exercises

• Exercises
• Solution proposals