Transcript L n-1

Probabilistic Computation with
DNA Molecules: The Probabilistic
Library model
Byoung-Tak Zhang
Summarized by HaYoung Jang
Introduction
Probabilistic library model
How to represent the joint probability of data
variables in DNA molecules (representation).
How to calculate conditional probabilities of
variables (inference).
How to update the probability distribution from
observed data (learning).
The Probablisitic Library Model
Computing Probabilities
Marginal probability of A
Marginal probability of B
Joint probability
Conditional probability
Updating Probabilty Distributions
1. Let the library L represent the current empirical distribution
P(X, Y)
2. Get a training example (x, y).
3. Classify x using L as described in the previous slide. Let this
class be y*
4. Update L
If y* = y, then Ln  Ln-1 + {Δc(u, v)} for u = x and v = y for (u, v) ∈
Ln-1
If y* ≠ y, then Ln  Ln-1 – {Δc(u, v)} for u = x and v ≠ y for (u, v) ∈ Ln1
5. Goto Step 2 if not terminated
Majority Function
Why does it work?
x1
x2
x3
y
0
0
0
0
0
0
1
0
0
1
0
0
0
1
1
1
1
0
0
0
1
0
1
1
1
1
0
1
1
1
1
1
Multiplexer