课程目录
- 001_welcome! (05 -35)_1
- 002_overview and motivation (19 -17)_1
- 003_distributions (04 -56)_1
- 004_factors (06 -40)_1
- 005_semantics & factorization (17 -20)_1
- 006_reasoning patterns (09 -59)_1
- 007_flow of probabilistic influence (14 -36)_1
- 008_conditional independence (12 -38)_1
- 009_independencies in bayesian networks (18 -18)_1
- 010_naive bayes (09 -52)_1
- 011_application - medical diagnosis (09 -19)_1
- 012_knowledge engineering example - samiam (14 -14
- 013_overview of template models (10 -55)_1
- 014_temporal models - dbns (23 -02)_1
- 015_temporal models - hmms (12 -01)_1
- 016_plate models (20 -08)_1
- 017_basic operations (13 -59)_1
- 018_moving data around (16 -07)_1
- 019_computing on data (13 -15)_1
- 020_plotting data (09 -38)_1
- 021_control statements - for while if statements
- 022_vectorization (13 -48)_1
- 023_working on and submitting programming exercise
- 024_overview - structured cpds (08 -00)_1
- 025_tree-structured cpds (14 -37)_1
- 026_independence of causal influence (13 -08)_1
- 027_continuous variables (13 -25)_1
- 028_pairwise markov networks (10 -59)_1
- 029_general gibbs distribution (15 -52)_1
- 030_conditional random fields (22 -22)_1
- 031_independencies in markov networks (04 -48)_1
- 032_i-maps and perfect maps (20 -59)_1
- 033_log-linear models (22 -08)_1
- 034_shared features in log-linear models (08 -28)_
- 035_knowledge engineering (23 -05)_1
- 036_overview - conditional probability queries (15
- 037_overview - map inference (09 -42)_1
- 038_variable elimination algorithm (16 -17)_1
- 039_complexity of variable elimination (12 -48)_1
- 040_graph-based perspective on variable eliminatio
- 041_finding elimination orderings (11 -58)_1
- 042_belief propagation (21 -21)_1
- 043_properties of cluster graphs (15 -00)_1
- 044_properties of belief propagation (9 -31)_1
- 045_clique tree algorithm - correctness (18 -23)_1
- 046_clique tree algorithm - computation (16 -18)_1
- 047_clique trees and independence (15 -21)_1
- 048_clique trees and ve (16 -17)_1
- 049_bp in practice (15 -38)_1
- 050_loopy bp and message decoding (21 -42)_1
- 051_max sum message passing (20 -27)
- 052_finding a map assignment (3 -57)
- 053_tractable map problems (15 -04)
- 054_dual decomposition - intuition (17 -46)
- 055_dual decomposition - algorithm (16 -16)
- 056_simple sampling (23 -37)
- 057_markov chain monte carlo (14 -18)
- 058_using a markov chain (15 -27)
- 059_gibbs sampling (19 -26)
- 060_metropolis hastings algorithm (27 -06)
- 061_inference in temporal models (19 -43)
- 062_inference - summary (12 -45)
- 063_maximum expected utility (25 -57)
- 064_utility functions (18 -15)
- 065_value of perfect information (17 -14)
- 066_regularization - the problem of overfitting
- 067_regularization - cost function (10 -10)
- 068_evaluating a hypothesis (07 -35)
- 069_model selection and train validation test sets
- 070_diagnosing bias vs variance (07 -42)
- 071_regularization and bias variance (11 -20)
- 072_learning - overview (15 -35)
- 073_maximum likelihood estimation (14 -59)
- 074_maximum likelihood estimation for bayesian net
- 075_bayesian estimation (15 -27)
- 076_bayesian prediction (13 -40)
- 077_bayesian estimation for bayesian networks (17
- 078_maximum likelihood for log-linear models (28 -
- 079_maximum likelihood for conditional random fiel
- 080_map estimation for mrfs and crfs (9 -59)
- 081_structure learning overview (5 -49)
- 082_likelihood scores (16 -49)
- 083_bic and asymptotic consistency (11 -26)
- 084_bayesian scores (20 -35)
- 085_learning tree structured networks (12 -05)
- 086_learning general graphs - heuristic search
- 087_learning general graphs - search and decomposa
- 088_learning with incomplete data - overview
- 089_expectation maximization - intro (16 -17)
- 090_analysis of em algorithm (11 -32)
- 091_em in practice (11 -17)
- 092_latent variables (22 -00)
- 093_summary - learning (20 -11)
- 094_class summary (24 -38)
课程介绍
Probabilistic graphical models (PGMs) are a rich framework for encoding probability distributions over complex domains: joint (multivariate) distributions over large numbers of random variables that interact with each other. These representations sit at the intersection of statistics and computer science, relying on concepts from probability theory, graph algorithms, machine learning, and more. They are the basis for the state-of-the-art methods in a wide variety of applications, such as medical diagnosis, image understanding, speech recognition, natural language processing, and many, many more. They are also a foundational tool in formulating many machine learning problems.
