@handdancin ...doesn't starts d-separate radio and petrol by his definition, and wouldn't that imply radio and petrol are independent if you know starts? I understand why they arent, but it seems to be at odds with the definition. Or does the vertical orientation matter or something?
@guharup Nope.. the inputs to such a machine would be through sensors.. so we will have sensors for an earthquake(on lines of richter scale) and also one for burglary(a camera).. Either of the sensors, upon occurence of corresponding activity will trigger the alarm. Burglary and earthquake are two independant activities being sensed independantly.
very low tech but still effective presentation! note how prof does NOT need to stand at a blackboard. Not even stand on an overhead projector! two camera angles (from top and front) do the trick! they have NOT centered the powerpoint slides (TRAVESTY for a institute of technology).
Not sure if it is low tech, this is exactly how lectures are conducted at USC electrical engineering. This is pretty much the standard way. There are smart boards etc but most profs find it too distracting.
I am not going to lie. There were times in this video where he gave me the stare of a rapist.
LeathaFace17 2 months ago
Nice lecture! First half is best.
Outline:
--------
1:00 - Types of things deducible via Joint Probability. Intro and motivation for Bayes networks
12:30 - How the complex joint prob distribution reduces in a Bayes network
20:30 - How benefits of a Bayes network depend on an appropriate ordering of variables
32:00 - How to use the Baye's network to infer about combinaions of events. Rules of thumb.
47:00 - lecture ends. Rest is background on IIT Kharagpur.
------
SiddharthBhaiVideos 4 months ago 8
Wow. This is amazing!
handdancin 4 months ago
@handdancin ...doesn't starts d-separate radio and petrol by his definition, and wouldn't that imply radio and petrol are independent if you know starts? I understand why they arent, but it seems to be at odds with the definition. Or does the vertical orientation matter or something?
handdancin 4 months ago
Very informative video on Bayesian Networks! Detailed examples are very useful.
StatisticsExpert 8 months ago
earthquake is dependant on burglary since unless you know of burglary you cant be sure whether alarm went off due to earthquake or burglary
guharup 1 year ago
@guharup Nope.. the inputs to such a machine would be through sensors.. so we will have sensors for an earthquake(on lines of richter scale) and also one for burglary(a camera).. Either of the sensors, upon occurence of corresponding activity will trigger the alarm. Burglary and earthquake are two independant activities being sensed independantly.
MsPayal1 11 months ago
@guharup
but then earthquake would be conditionally independent of burglary, given alarm.
mauroprovatos 4 months ago
Excellent lecture. Very useful
kordusit 1 year ago
Comment removed
aajoshi1 1 year ago
veryyyyy informative....must say i am impressed...helped me for my final exam!...thanks Prof. P. Dasgupta,
MrFreshShizzle 1 year ago
well done this is what youtube should be used for
andrewzot 1 year ago 4
very low tech but still effective presentation! note how prof does NOT need to stand at a blackboard. Not even stand on an overhead projector! two camera angles (from top and front) do the trick! they have NOT centered the powerpoint slides (TRAVESTY for a institute of technology).
BTW this extended example is from J Pearl.
kalm77 2 years ago
@kalm77
Not sure if it is low tech, this is exactly how lectures are conducted at USC electrical engineering. This is pretty much the standard way. There are smart boards etc but most profs find it too distracting.
aajoshi1 1 year ago
Comment removed
prognanica 2 years ago
better than my georgia tech lectures.. wish I found it sooner.. at least i get to use it for finals.
celinesrivera 2 years ago
In 22:00, why is Earthquake dependent on Burglary?
Archangel10123 2 years ago
Yeah. I think it is to do with the "chain of inference" the prof has been going on about at least twice prior to 22:00. The latest at 19:30.
SO with the a priori variables interchanged, new dependencies arise from the "CHAIN" of inference.
I too fell asleep by the way. He has very relaxing voice!
kalm77 2 years ago
This has been flagged as spam show
Nice work. keep it up. mean time come for social media marketing for esteembpo**com
lawerenceburch 2 years ago
better than my real lecture.
Anyone else get reminded of the old OU progs on BBC?
jamesj629 2 years ago
Yeah, that was very clear and surprisingly interesting. Thanks! Also, nice handwriting.
neurotrash84 2 years ago 16
Brilliant .. and very well presented .. thank you very much.
adelpierro 2 years ago 11
That was very useful! Thank you.
crossmynameout 2 years ago 3
Nice lecture course and good introduction to bayesian
videobob8 3 years ago 2