probabilistic, reasoning, artificial, computer, intelligence, IOE, Sushant, Pulchowk, AI,
Statistical techniques used in practical data analysis. e.g. t-tests, ANOVA, regression, correlation;
The use of probabilistic models in psychology and linguistics
Machine learning and computational linguistics/NLP
Measure theory (in fact, almost anything involving infinite sets)
Using logic and probability to handle uncertain situation
Probability based reasoning is same as understanding directly from the knowledge that a given probability rating based on uncertainty present
2. • Statistical techniques used in practical data analysis.
e.g. t-tests, ANOVA, regression, correlation;
• The use of probabilistic models in psychology and linguistics
• Machine learning and computational linguistics/NLP
• Measure theory (in fact, almost anything involving infinite sets)
PROBABILITY& STATISTICS
3. • Using logic andprobabilityto handleuncertainsituation
• Probabilitybasedreasoning is same as understanding
directlyfrom the knowledgethat agiven probability
rating basedon uncertaintypresent
PROBABILITYREASONING
4. • Crossing the street in traffic.
We’ve all done this:
UNCERTAINTY AND UNCERTAIN REASONING : Intuition warmup
you’re in a hurry, so instead of waiting for the “walk” sign
you look both ways and see that the nearest cars are far
enough away that you can cross safely before they arrive
where you are. You start walking and (I’m guessing) make
it across just fine.
5. • The cop and the man in the
window.
UNCERTAINTY AND UNCERTAIN REASONING : Intuition warmup
You’re a police officer out on patrol late at night. You
hear an alarm go off and follow the sound to a jewelry
store. When you arrive, you see a broken window and a
man crawling out of it wearing black clothes and a mask,
carrying a sack which turns out to be full of jewelry.
6. • Medical diagnosis
1. the person has a cold (h1),
2. lung disease (h2),
3. heartburn (h3).
4. tuberculosis (h4)
UNCERTAINTY AND UNCERTAIN REASONING : Intuition warmup
Suppose we observe a person coughing, and we
consider three hypotheses as explanations.
7. • Doing science requires the ability to cope with uncertainty
UNCERTAINTY
• Human familiar deductive logic is great for reasoning
• With science & machines, we need a procedure for
determining which conclusions to draw
This should take the form of an inductive logic.
30. • Probabilistic reasoningare used when outputor outcomes are
unpredictable
• One way to express confidenceabout such event is probability
• Probabilityof an uncertainevent ismeasure of degree of likelihood
of occurrence of that event
PROBABILITYREASONING : HANDLING UNCERTAINTY
31.
32.
33. To aid in the
interpretation of
gene lists,
PheNetic was
built on top of
ProbLog.
ProbLog is used
to reason over
heterogeneous
data sources like
the Helsinki
Biomine
database.
38. UNCERTAINTY
“All swans are white”, a universal generalization
from observation of many, many white swans
This was before Europeans
went to Australia. When they
got there, they discovered that
Australian swans are black.
D’oh!
Early modern philosophy in Europe
40. NEEDOF PROBABILISTIC REASONING
Cognitive sciences
(e.g. linguistics, psychology, AI, philosophy of
mind & epistemology):
we’re trying to understand human intelligence,
using noisy and uncertain information
to make (hopefully) reasonable/ intelligent decision