10.3 Kohonen Self-Organizing Map

[@radsimu]

very interesting. but what is the activation function for performing classification of new inputs with this thing? It is something that encourages small distances between weights and inputs, which is very different from the classic weighted sum of some activation function what you normally see in neural networks. Which makes me wonder why are we even calling the nodes/centroids things "neurons"

[@deebo7276]

Great presentation for the fact you point out and clarify that the initial allocation of clusters to 2D nodes is not important, and that the 2D node proximity to neighboring nodes is strictly a result of the neighborhood learning. You say in some cases clusters are allocated to 2D nodes (prior to learning I assume?), could you say anything about this? Is the point of that to speed up the learning as n-dimensional nodes will already have closer proximities as indicated by the 2D map, and don't require as many iterations?

[@hasanmashat6306]

Great explanation charu!

[@km-ss7st]

Very helfful explaination! Thankyou

[@glaswasser]

How do we get an input vector containing information about the genres? Will it just contain numeric values for "Drama", "music", "arts" and "music", with a 1 or a 0 depending if the input belongs to the class or not, or how does that work?

[@charuaggarwal2740]

The weight vector W_i is like the centroid of k-means. It can be considered a numeric description of the cluster, and can also be considered a prototype. The dimensionality of this vector is the same as that of the data set.

[@dr.rajanishkumarkaushal9012]

Nothing showing on slides

[@AndyD6]

are you blind sir, i see slides fine on video. very informative good teaching

[@piyushwagh7960]

can u prrovide python code link?

[@nex7886]

thank you sir, very helpful good explanation

[@AndyD6]

very thank for this video, i use for my study in degree computer science thank you prof

[@nex7886]

i, too am using this for class computer science degree in studying. i like this teaching videa very good thank you prof

[@AndyD6]

@@nex7886 i passed!

[@subramaniamsrivatsa2719]

very beautiful and succintly explained.. Thank you!

[@sshiv908]

nice slides but it would have been better if u uses diagrams to explain it

[@dr.rajanishkumarkaushal9012]

R u teaching someone are reading self

[@AndyD6]

he is teaching to all of us who watch video, like i am. good learning

[@vallabhchugh2075]

good but not for starters

[@jackjohnsen9719]

Thanks for the video. You might want to slow down a bit - you are speaking too fast and it gets pretty hard to understand. Anyway, I still managed to learn something, so thanks :)

[@AndyD6]

i have tip for you sir, you can set speed to 0.75x or 0.5x or 0.25x via video interface. i use this plenty time for learning. i am very fast brain so i set video to 2x speed and i learn twice as fast haha

[@glaswasser]

I watched at 1.25 and it was fine for me - so it always depends if the topic is completely new to you or not I think