Sam Roweis

1
Sam Roweis

• Machine learning

2
Research

• What kinds of problems am I interested in?
• intelligent processing of sensory informationfor
  classification, compression, denoising,visualizat
  ion, synthesis
• automatic structure discovery in data
• designing algorithms to do all this
• tools statistical method, probabilistic models,
  linear algebra, approximate solutions

3
Data

• What kinds of data do I work with?
• scientific/industrial data (astrophysics/plants)
• speech, audio, EEG, other time series
• text/web data
• very large datasets in very high dimensions

4
Current Projects Applications

• Nonlinear Dimensionality Reduction
• LLE algorithm(image libraries SDDS)
• Speech and Audio Processing
• CASA, noise reduction, audio textures(Intelligent
  hearing aids auto-diary)
• Text/Web Analysis
• probabilistic models, continuous
  variables(Whizbang! Labs)

5
(No Transcript)
6
Courses

• CSC2515 - Machine Learning
• Wednesdays 3pm-5pm
• basics of machine learningclassifcation,
  regression, clusteringsequence analysis, novelty
  detection
• naïve Bayes, logistic regression, k-means,
  mixtures of experts, HMMs, decision trees,
  support vector machines, PCA, ICA, parameter
  optimization,EM algorithm