Teaching : Statistical Machine Learning and Data Mining (MS1b HT2014)

Term:	Hilary Term, January 20 - March 14, 2014
Lecturer:	Yee Whye Teh
Teaching Assistants:	Part C: Thibaut Lienart MSc: Balaji Lakshminarayanan and Maria Lomeli
Lectures:	1400-1500 Mondays (Math Institute L4) 1000-1100 Wednesdays (Math Institute L3)
Problem Sheets: Part C Classes: MSc Classes:	Due 1200 the Fridays prior to classes in 1 South Parks Road (SPR). 1600-1700 Tuesdays (Weeks 3-8) in 1 SPR Seminar Room. 1400-1600 Tuesdays (Weeks 3, 5, 7, 9) in 2 SPR Seminar Room.
Practicals: (MSc only)	1400-1800 Friday Week 5 (unassessed) in 1 SPR Computing Lab. 1400-1800 Friday Week 7 (assessed) in 1 SPR Computing Lab.
Google Group:	https://groups.google.com/forum/?hl=en-GB#!forum/smldm

News

• Welcome to Statistical Machine Learning and Data Mining!
• Please sign up at the Google Groups above. This is for communications with lecturer and TAs. You are also encouraged to ask questions on the group. The lecturer and TAs will answer when they can, and other students can answer as well.
• Cheat sheets for calculus and algebra for matrices and vectors:
  • Matrix and Gaussian identities - short useful reference for machine learning.
  • Linear Algebra Review and Reference - useful selection for machine learning.
  • The Matrix Cookbook - extensive reference.

Slides and Problem Sheets

• All slides. 4 slides per page.
• R Demos.
  
  
• Week 1 Slides. 4 slides per page.
• Week 2 Slides. 4 slides per page.
• Week 3 Slides. 4 slides per page.
• Week 4 Slides. 4 slides per page.
• Week 5 Slides. 4 slides per page.
• Week 6 Slides. 4 slides per page.
• Week 7 Slides. 4 slides per page.
• Week 8 Slides. 4 slides per page.
  
  
• Part C Problem sheet 1. Solution.
• Part C Problem sheet 2. Solution.
• Part C Problem sheet 3. Solution.
• Part C Problem sheet 4. Solution.
• Part C Problem sheet 5. Solution.
• Part C Problem sheet 6. Solution. Data and script.
  
  
• MSc Problem sheet 1. Solution.
• MSc Problem sheet 2. Solution.
• MSc Problem sheet 3. Solution.
• MSc Problem sheet 4. Solution.
  
  
• MSc Week 5 Practical. Solution.
  Data: X covariates. Y age.
• MSc Week 7 Practical. Data and script. Solution.
  

R

You will need to use the R statistical programming language and environment for this course.

• An introduction to R.
  [pdf]
• If you do not have access to a computer with R already installed, you can download it at
  http://cran.r-project.org/.
• If you happen to use Emacs, we recommend for interaction with R the
  ESS (Emacs speaks Statistics) package.

Textbooks

Recommended textbooks on statistical machine learning and data mining. The technical developments follow Hastie et al most closely, while programming follows mostly Ripley. Various other sources are used as well.

• Ripley, Pattern Recognition and Neural Networks, Cambridge University Press.
• Hastie, Tibshirani and Friedman, The Elements of Statistical Learning, Springer-Verlag.
  [ebook]
• James, Witten, Hastie and Tibshirani, An Introduction to Statistical Learning with Applications in R, Springer-Verlag.
  [ebook]
• Bishop, Pattern Recognition and Machine Learning, Springer.
• Duda, Hart and Stork, Pattern Classification, Wiley-Interscience.

Machine Learning and Data Mining in the News

• 8 Jan 2014: Nature: Computer science: The learning machines
  Article about recent developments in machine learning (particularly deep learning) and how it has caught the attention of the scientific community and industry. Reply by Yann LeCun.
• 14 Jan 2014: Bloomberg: Meet Facebook's Head of Artificial Intelligence
  Interview with Yann LeCun, a leading figure in machine learning, particularly in deep neural networks and computer vision.
• 16 Jan 2014: Wired: Meet the Man Google Hired to Make AI a Reality
  A brief history of deep learning and its leading proponent Geoff Hinton.
• 22 Jan 2014: Wired: How a Math Genius Hacked OkCupid to Find True Love
  How to do online dating successfully with machine learning :)
• 27 Jan 2014: Google buys UK artificial intelligence start-up DeepMind
  DeepMind is a London-based start-up that develops machine learning techniques (deep learning, reinforcement learning, Bayesian learning) for AI and computer games.
• 28 Jan 2014: Deep Learning: Teaching Computers To Think Like People
  Interview with Philip Resnik, Geoffrey Hinton and Richard Socher.
• 13 Feb 2014: Enslave the robots and free the poor
  Opinion piece in Financial Times. General interest about how intelligent machines will impact the economy and society.
• 13 Feb 2014: Netflix Is Building an Artificial Brain Using Amazon’s Cloud
  Netflix movie recommendation using deep learning.
• 15 Feb 2014: The Dawn of the Age of Artificial Intelligence
  Not exactly about machine learning, but it does underpin a lot of the technologies.
• 26 Feb 2014: Reddit Machine Learning
  Just found out there's a reddit thread on machine learning, and currently Yoshua Bengio is doing an AMA.

Schedule