I'm very glad to have found this series. I was "assigned" to read Deep Learning more than a year ago and haven't been able to bring myself to read it. . . but I can sit in front of a video and hear/see the subjects. It's clear this production is a courtesy & is very thoughtful for you to have recorded, edited, & uploaded it with zero ego content. So thank you very much. Cheers from California.
lo., the transpose of matrix is super simple, first get the size of matrix by reversing the m*n to n*m..then write all the rows as columns and you have a transposed matrix.
Hahaha this is gold! I get people are upset at the lack of rigor but I feel like it’s a BIG ask to succinctly cover all this linear algebra in 50 mins. Gotta sacrifice depth for speed. “The determine method taught in textbooks is a total pain in the ass” - agreed lol
22:21 He didn't give precise answer to the question asked here. Explanation that i have for this question is that, vector has two things 1) magnitude 2) direction. When we write them as just list of numbers , the point of reference is assumed to be zero(origin) so your direction is [1 2 3] - [0 0 0] = [1 2 3]. so in that interpretation it is a vector but if you don't assume a reference point then that can't represent vector in true sense and it is just a static point lying in the space that has no direction attached to it and its length is 0. If instead of 0 one chooses to use different point of implicit reference then the same vector will be represented using different numbers in that list but it would be the same as [1 2 3] , its just that its head and tails have changed and now it is no longer ending at that static point. For defining vector you need to tell its head and tail point and usually tail is assumed to be at zero(origin) and hence people are confused between point and vectors. I know this from a multivariable calculus course here on youtube from harvard, so think of vectors as some transformation of whole space in which you are moving all the points in the direction given by the vector by a distance that equals its length but point is just a static entity.
Thanks for uploading. For next time, would you mind moving the camera accordingly, it was quite hard to follow what he was writing on the board. Thanks.
audio is crap and explanations are unclear. Hard to get any useful info out of this video. The concepts are simple but the notations are often confusing and this lecture has not done a good job on explaining neither the concept nor the notations
![ARGENTINA vs. CHILE [3-0] | RESUMEN | ELIMINATORIAS SUDAMERICANAS | FECHA 7](http://i.ytimg.com/vi/De3AWmXzhuA/mqdefault.jpg)
I'm very glad to have found this series. I was "assigned" to read Deep Learning more than a year ago and haven't been able to bring myself to read it. . . but I can sit in front of a video and hear/see the subjects. It's clear this production is a courtesy & is very thoughtful for you to have recorded, edited, & uploaded it with zero ego content. So thank you very much. Cheers from California.
Very nice discussion! I sometimes get lost on detailed explanations so this helped me to better understand the purpose of each thing.
lo., the transpose of matrix is super simple, first get the size of matrix by reversing the m*n to n*m..then write all the rows as columns and you have a transposed matrix.
Hahaha this is gold! I get people are upset at the lack of rigor but I feel like it’s a BIG ask to succinctly cover all this linear algebra in 50 mins. Gotta sacrifice depth for speed. “The determine method taught in textbooks is a total pain in the ass” - agreed lol
22:21 He didn't give precise answer to the question asked here.
Explanation that i have for this question is that, vector has two things 1) magnitude 2) direction.
When we write them as just list of numbers , the point of reference is assumed to be zero(origin) so your direction is [1 2 3] - [0 0 0] = [1 2 3]. so in that interpretation it is a vector but if you don't assume a reference point then that can't represent vector in true sense and it is just a static point lying in the space that has no direction attached to it and its length is 0.
If instead of 0 one chooses to use different point of implicit reference then the same vector will be represented using different numbers in that list but it would be the same as [1 2 3] , its just that its head and tails have changed and now it is no longer ending at that static point.
For defining vector you need to tell its head and tail point and usually tail is assumed to be at zero(origin) and hence people are confused between point and vectors. I know this from a multivariable calculus course here on youtube from harvard, so think of vectors as some transformation of whole space in which you are moving all the points in the direction given by the vector by a distance that equals its length but point is just a static entity.
That's the physic's interpretation of a vector. The real definition is more abstract.
thank you for sharing, Can you please upload another lecture on the same topic with different instructor.
lol
😂😂😂
Thanks for uploading. For next time, would you mind moving the camera accordingly, it was quite hard to follow what he was writing on the board. Thanks.
Sure thing! I'll try my best
15:00, Matrix-product is not the same as Dot-product. Dot-product aka. scalar-product can only be done between Vectors.
in numpy it is still referred as `np.ndarray.dot`
You can actually think of matrix product of 2 matrices A and B as the dot product of the ith row of A, and the Jth column of B.
that's worst lecture of Linear algebra :D :D, Gavin Crooks is being too accurate HAHAHAHAHHAHAHA
Unfortunately, sound quality/volume of the video series on the DL book is very low, making it a pain to follow.
@tagoma that is unfortunate! this video should have decent sound. I just checked on 2 devices and it works ok for me :/ are you on a mac?
Thanks for replying. I am using a PC. A laptop. Volume of the "Deep Learning Book" video series is way lower than other videos.
ugh! sorry about that. I'll try to improve in future edits. Unfortunately cranking up audio all the way is the only suggestion i got :/
yes,, I have to up the volume to max to follow
harsh treatment of roger penrose there
Sir Roger Penrose 🎩🎩
14:10 LMAO should be (5,4) not (5,5)
What he say about rank at @7:15? I understand it.
Gavin Crooks?Is he the Crooks in Crooks Fluctuation Theorem?
yes
audio is crap and explanations are unclear. Hard to get any useful info out of this video. The concepts are simple but the notations are often confusing and this lecture has not done a good job on explaining neither the concept nor the notations
There are a lot of resources for learning this subject such as 3blue1brown. This series is merely a companion to the book called Deep Learning.
He shouldn't teach, just saying...
This guy is not fit enough to clean my toilets