CS 3750  Advanced Topics in Machine Learning (ISSP 3535)

Time:  Tuesday, Thursday 11:00am - 12:15pm  
Location: Sennott Square, Room 5313

Instructor:  Milos Hauskrecht
Computer Science Department
5329 Sennott Square
phone: x4-8845
e-mail: milos_at_cs_pitt_edu
office hours: TBA

Background: math, algebra, and probabilities


Matrices, Algebra

Density estimation

Graphical models: BBNs and MRFs

Grahical models: Bayesian Belief Networks, Markov Random Fields, Conversion of BBNs to MRFs, Clique trees (tree decompositions of the MRFs)

Exact inference algorithms: factor graphs

Exact inference algorithms on clique trees

Monte Carlo methods:

Variational methods:

Loopy Belief propagation (not covered in Fall 2018)

Learning graphical models from data (not covered in Fall 2018)

Learning BBNs:

Learning MRFs:

Component analysis models




Applications of PCA/SVD:

Non-negative matrix factorization:

Latent Variable Component Analysis Models

Probabilistic PCA

Other latent variable models

Non-linear dimensionality reduction and kernels

Laplacian Eigenmaps for dimensionality reduction

Kernel PCA

Diffusion kernels (not covered in 2018)

Models for document analysis, information retrieval and link analysis

Probabilistic latent semantic analysis (pLSA)

pLSA for link analysis

Latent Dirichlet Allocation

Word and word similarity models: word2vec, CBOW, graph-based models

Semi-supervised learning, label propagation (not covered in Fall 2018)

Gaussian Processes

Time series and sequence analysis models

Markov models, Hidden Markov models, Dynamic Belief networks

Linear Dynamical model

Approximate inference in Dynamic belief networks: Particle filtering

Autoregressive models (first and k-order AR, ARMA, ARIMA)

Deep learning

Recurrent Neural Networks, LSTM

Convolutional neural networks

Deep generative models

Generative Adversarial Networks

Deep learning: Applications


Last updated by milos on 11/19/2018