Probabilistic Models  Models describe how (a portion of) the world works  Models are always simplifications  May not account for every variable  May not account for all interactions between variables  “All models are wrong; but some are useful.” – George E. P. Box  What do we do with probabilistic models?  We (or our agents) need to reason about unknown variables, given evidence  Example: explanation (diagnostic reasoning)  Example: prediction (causal reasoning)  Example: value of information 1 This slide deck courtesy of Dan Klein at UC Berkeley

Probabilistic Models  A probabilistic model is a joint distribution over a set of variables  Inference: given a joint distribution, we can reason about unobserved variables given observations (evidence)  General form of a query:  This conditional distribution is called a posterior distribution or the the belief function of an agent which uses this model Stuff you care about Stuff you already know 2

Probabilistic Inference  Probabilistic inference: compute a desired probability from other known probabilities (e.g. conditional from joint)  We generally compute conditional probabilities  P(on time | no reported accidents) = 0.90  These represent the agent’s beliefs given the evidence  Probabilities change with new evidence:  P(on time | no accidents, 5 a.m.) = 0.95  P(on time | no accidents, 5 a.m., raining) = 0.80  Observing new evidence causes beliefs to be updated 3

The Product Rule  Sometimes have conditional distributions but want the joint  Example: RP sun0.8 rain0.2 DWP wetsun0.1 drysun0.9 wetrain0.7 dryrain0.3 DWP wetsun0.08 drysun0.72 wetrain0.14 dryrain0.06 4

The Chain Rule  More generally, can always write any joint distribution as an incremental product of conditional distributions 5

Bayes’ Rule  Two ways to factor a joint distribution over two variables:  Dividing, we get:  Why is this at all helpful?  Lets us build one conditional from its reverse  Often one conditional is tricky but the other one is simple  Foundation of many systems we’ll see later  In the running for most important AI equation! That’s my rule! 6

Inference with Bayes’ Rule  Example: Diagnostic probability from causal probability:  Example:  m is meningitis, s is stiff neck  Note: posterior probability of meningitis still very small  Note: you should still get stiff necks checked out! Why? Exampl e givens 7

Ghostbusters, Revisited  Let’s say we have two distributions:  Prior distribution over ghost location: P(G)  Let’s say this is uniform  Sensor reading model: P(R | G)  Given: we know what our sensors do  R = reading color measured at (1,1)  E.g. P(R = yellow | G=(1,1)) = 0.1  We can calculate the posterior distribution P(G|r) over ghost locations given a reading using Bayes’ rule:

Independence  Two variables are independent in a joint distribution if:  Says the joint distribution factors into a product of two simple ones  Usually variables aren’t independent!  Can use independence as a modeling assumption  Independence can be a simplifying assumption  Empirical joint distributions: at best “close” to independent  What could we assume for {Weather, Traffic, Cavity}? 9

Example: Independence? TWP warmsun0.4 warmrain0.1 coldsun0.2 coldrain0.3 TWP warmsun0.3 warmrain0.2 coldsun0.3 coldrain0.2 TP warm0.5 cold0.5 WP sun0.6 rain0.4 10

Example: Independence  N fair, independent coin flips: H0.5 T H T H T 11

Conditional Independence  P(Toothache, Cavity, Catch)  If I have a cavity, the probability that the probe catches in it doesn't depend on whether I have a toothache:  P(+catch | +toothache, +cavity) = P(+catch | +cavity)  The same independence holds if I don’t have a cavity:  P(+catch | +toothache,  cavity) = P(+catch|  cavity)  Catch is conditionally independent of Toothache given Cavity:  P(Catch | Toothache, Cavity) = P(Catch | Cavity)  Equivalent statements:  P(Toothache | Catch, Cavity) = P(Toothache | Cavity)  P(Toothache, Catch | Cavity) = P(Toothache | Cavity) P(Catch | Cavity)  One can be derived from the other easily 12

Conditional Independence  Unconditional (absolute) independence is very rare (why?)  Conditional independence is our most basic and robust form of knowledge about uncertain environments:  What about this domain:  Traffic  Umbrella  Raining  What about fire, smoke, alarm? 13

Bayes’ Nets: Big Picture  Two problems with using full joint distribution tables as our probabilistic models:  Unless there are only a few variables, the joint is WAY too big to represent explicitly  Hard to learn (estimate) anything empirically about more than a few variables at a time  Bayes’ nets: a technique for describing complex joint distributions (models) using simple, local distributions (conditional probabilities)  More properly called graphical models  We describe how variables locally interact  Local interactions chain together to give global, indirect interactions 14

Example Bayes’ Net: Insurance 15

Example Bayes’ Net: Car 16

Graphical Model Notation  Nodes: variables (with domains)  Can be assigned (observed) or unassigned (unobserved)  Arcs: interactions  Indicate “direct influence” between variables  Formally: encode conditional independence (more later)  For now: imagine that arrows mean direct causation (in general, they don’t!) 17

Example: Coin Flips X1X1 X2X2 XnXn  N independent coin flips  No interactions between variables: absolute independence 18

Example: Traffic  Variables:  R: It rains  T: There is traffic  Model 1: independence  Model 2: rain causes traffic  Why is an agent using model 2 better? R T 19

Example: Traffic II  Let’s build a causal graphical model  Variables  T: Traffic  R: It rains  L: Low pressure  D: Roof drips  B: Ballgame  C: Cavity 20

Example: Alarm Network  Variables  B: Burglary  A: Alarm goes off  M: Mary calls  J: John calls  E: Earthquake! 21

Bayes’ Net Semantics  Let’s formalize the semantics of a Bayes’ net  A set of nodes, one per variable X  A directed, acyclic graph  A conditional distribution for each node  A collection of distributions over X, one for each combination of parents’ values  CPT: conditional probability table  Description of a noisy “causal” process A1A1 X AnAn A Bayes net = Topology (graph) + Local Conditional Probabilities 22

Probabilities in BNs  Bayes’ nets implicitly encode joint distributions  As a product of local conditional distributions  To see what probability a BN gives to a full assignment, multiply all the relevant conditionals together:  Example:  This lets us reconstruct any entry of the full joint  Not every BN can represent every joint distribution  The topology enforces certain conditional independencies 23

Example: Coin Flips h0.5 t h t h t X1X1 X2X2 XnXn Only distributions whose variables are absolutely independent can be represented by a Bayes’ net with no arcs. 24

Example: Traffic R T +r1/4 rr 3/4 +r +t3/4 tt 1/4 rr +t1/2 tt 25

Example: Alarm Network Burglary Earthqk Alarm John calls Mary calls BP(B) +b0.001 bb EP(E) +e0.002 ee BEAP(A|B,E) +b+e+a0.95 +b+e aa b ee +a0.94 +b ee aa 0.06 bb +e+a0.29 bb +e aa 0.71 bb ee +a0.001 bb ee aa AJP(J|A) +a+j0.9 +a jj 0.1 aa +j0.05 aa jj 0.95 AMP(M|A) +a+m0.7 +a mm 0.3 aa +m0.01 aa mm 0.99

Bayes’ Nets  A Bayes’ net is an efficient encoding of a probabilistic model of a domain  Questions we can ask:  Inference: given a fixed BN, what is P(X | e)?  Representation: given a BN graph, what kinds of distributions can it encode?  Modeling: what BN is most appropriate for a given domain? 27

Building the (Entire) Joint  We can take a Bayes’ net and build any entry from the full joint distribution it encodes  Typically, there’s no reason to build ALL of it  We build what we need on the fly  To emphasize: every BN over a domain implicitly defines a joint distribution over that domain, specified by local probabilities and graph structure 28

Size of a Bayes’ Net  How big is a joint distribution over N Boolean variables? 2N2N  How big is an N-node net if nodes have up to k parents? O(N * 2 k+1 )  Both give you the power to calculate  BNs: Huge space savings!  Also easier to elicit local CPTs  Also turns out to be faster to answer queries 29

Example: Independence  For this graph, you can fiddle with  (the CPTs) all you want, but you won’t be able to represent any distribution in which the flips are dependent! h0.5 t h t X1X1 X2X2 All distributions 30

Topology Limits Distributions  Given some graph topology G, only certain joint distributions can be encoded  The graph structure guarantees certain (conditional) independences  (There might be more independence)  Adding arcs increases the set of distributions, but has several costs  Full conditioning can encode any distribution X Y Z X Y ZX Y Z 31

Independence in a BN  Important question about a BN:  Are two nodes independent given certain evidence?  If yes, can prove using algebra (tedious in general)  If no, can prove with a counter example  Example:  Question: are X and Z necessarily independent?  Answer: no. Example: low pressure causes rain, which causes traffic.  X can influence Z, Z can influence X (via Y)  Addendum: they could be independent: how? XYZ

Causal Chains  This configuration is a “causal chain”  Is X independent of Z given Y?  Evidence along the chain “blocks” the influence XYZ Yes! X: Low pressure Y: Rain Z: Traffic 33

Common Cause  Another basic configuration: two effects of the same cause  Are X and Z independent?  Are X and Z independent given Y?  Observing the cause blocks influence between effects. X Y Z Yes! Y: Project due X: Newsgroup busy Z: Lab full 34

Common Effect  Last configuration: two causes of one effect (v-structures)  Are X and Z independent?  Yes: the ballgame and the rain cause traffic, but they are not correlated  Still need to prove they must be (try it!)  Are X and Z independent given Y?  No: seeing traffic puts the rain and the ballgame in competition as explanation?  This is backwards from the other cases  Observing an effect activates influence between possible causes. X Y Z X: Raining Z: Ballgame Y: Traffic 35

The General Case  Any complex example can be analyzed using these three canonical cases  General question: in a given BN, are two variables independent (given evidence)?  Solution: analyze the graph 36

Example  Variables:  R: Raining  T: Traffic  D: Roof drips  S: I’m sad  Questions: T S D R Yes 37

Causality?  When Bayes’ nets reflect the true causal patterns:  Often simpler (nodes have fewer parents)  Often easier to think about  Often easier to elicit from experts  BNs need not actually be causal  Sometimes no causal net exists over the domain  E.g. consider the variables Traffic and Drips  End up with arrows that reflect correlation, not causation  What do the arrows really mean?  Topology may happen to encode causal structure  Topology only guaranteed to encode conditional independence 38

Example: Traffic  Basic traffic net  Let’s multiply out the joint R T r1/4 rr 3/4 r t tt 1/4 rr t1/2 tt r t3/16 r tt 1/16 rr t6/16 rr tt 39

Example: Reverse Traffic  Reverse causality? T R t9/16 tt 7/16 t r1/3 rr 2/3 tt r1/7 rr 6/7 r t3/16 r tt 1/16 rr t6/16 rr tt 40

Example: Coins  Extra arcs don’t prevent representing independence, just allow non-independence h0.5 t h t X1X1 X2X2 h t h | h0.5 t | h0.5 X1X1 X2X2 h | t0.5 t | t  Adding unneeded arcs isn’t wrong, it’s just inefficient

Changing Bayes’ Net Structure  The same joint distribution can be encoded in many different Bayes’ nets  Causal structure tends to be the simplest  Analysis question: given some edges, what other edges do you need to add?  One answer: fully connect the graph  Better answer: don’t make any false conditional independence assumptions 42

Example: Alternate Alarm 43 Burglary Earthqua ke Alarm John calls Mary calls John calls Mary calls Alarm Burglary Earthqua ke If we reverse the edges, we make different conditional independence assumptions To capture the same joint distribution, we have to add more edges to the graph

Bayes’ Nets  Bayes’ net encodes a joint distribution  How to answer queries about that distribution  Key idea: conditional independence  How to answer numerical queries (inference)  (More later in the course) 44