I didn't see the SuperDataScience username at first, but this needs to make its way into the Udemy Deep Learning course with Kirill and Hadelin. The explanation of dimensionality reduction as shown in this video is not in the course from what I can tell. It would also help to create a 3D visualization that could be rotated to show clusters of nodes. I didn't get it until I saw this video, so thank you.
Also, based on your explanation, it seems that images like the ones on this page (www.ai-junkie.com/ann/som/som3.html - referenced in the Udemy course) are misleading because the width of the neighborhood is not dictated by the 2D grid but by the location of the neurons in higher dimensional space, correct?
It's fascinating stuff! BUT, please don't record a video stoned. Rule of thumb: if one sounds stoned at 1.25x speed, one definitely sounds boring. Please improve and keep churning out good content!
For a channel titled "Art of Visualization" I honestly expected a more...-visual explanation. Aside from that, great explanation.
these work wonders for emotional capacity for conscious AI 💯😏
I didn't see the SuperDataScience username at first, but this needs to make its way into the Udemy Deep Learning course with Kirill and Hadelin. The explanation of dimensionality reduction as shown in this video is not in the course from what I can tell. It would also help to create a 3D visualization that could be rotated to show clusters of nodes. I didn't get it until I saw this video, so thank you.
Also, based on your explanation, it seems that images like the ones on this page (www.ai-junkie.com/ann/som/som3.html - referenced in the Udemy course) are misleading because the width of the neighborhood is not dictated by the 2D grid but by the location of the neurons in higher dimensional space, correct?
Can you share the code? also, describe the hyperparameters?
Great explanation
Thanks for your feedback Rishik!
How did we compute the lateral distance?
Great Work, Thanks a lot
can you please share the codes? thanks
I'm looking for code that compress the data
It's fascinating stuff! BUT, please don't record a video stoned. Rule of thumb: if one sounds stoned at 1.25x speed, one definitely sounds boring. Please improve and keep churning out good content!
:D
I will tell Bruno your opinion :D
Thank you :)
😂😂
how to save and test on external data?
Thank u
can you share the code ?
only forward evaluations in evolution neural networks
if you in any way label, you dont have an unsupervised analysis
therefore soms are not unsupervised, if you have a goodness value
fake algorithm
like you are doing some sort of pca algorithm by linearly fitting the weight as the data x average
what you are fitting in the data, when you should not fit anything, only similarity
Where is the map? XD
Did you watch till the end?))
Mateus Maia, 6:06
My Internet bugged lol
My training took 0.07 seconds
kinda useless to explain SOM's and then use a pre-built library