MLSS-Indo-VAE-GAN
From Lia Anggraini to Everyone: (9:57 AM)
a bit sleepy but feeling awesome for today's lecture! :D its NLP
From Operator CLOVE 10 to Everyone: (9:57 AM)
:D
From Georgios to Everyone: (10:00 AM)
Good morning everyone from the UK.
From Tisa Siti Saadah to Everyone: (10:00 AM)
fine
From Mohd to Everyone: (10:00 AM)
Morning
From Tisa Siti Saadah to Everyone: (10:00 AM)
morning
From Robby Hardi to Everyone: (10:00 AM)
morning
From Paola to Everyone: (10:00 AM)
Hi :)
From Ebuka Oguchi to Everyone: (10:00 AM)
Good Morning
From Ayuni Mohamed to Everyone: (10:00 AM)
good morning from Malaysia
From Mohd to Everyone: (10:01 AM)
Ended at 11pm for me last night
From Operator CLOVE 10 (Ade) to Everyone: (10:04 AM)
you can give questions on rocket chat, #lecture11 channel
From Operator CLOVE 10 (Ade) to Everyone: (10:15 AM)
apologize for a technical glitch
From Operator CLOVE 10 (Ade) to Everyone: (11:05 AM)
Hello all! I think Katharina already join here, so please write your questions on the rocket chat!
From Operator CLOVE 10 (Ade) to Everyone: (11:19 AM)
Hi all, please feel free to give questions to Katharina
From Dedy Rahman Wijaya to Everyone: (11:28 AM)
Can be CNN and RNN combined for NLP?
From Sabrina to Everyone: (11:30 AM)
yes
From Wawan Cenggoro to Everyone: (11:30 AM)
yes we can
From Rian Adam Rajagede, S.Kom., M.Cs. to Everyone: (11:30 AM)
yes we can
From Maryam Fatima to Everyone: (11:48 AM)
I really like how energetic the lecturer is. :) Thanks Katharine
From Ebuka Oguchi to Everyone: (11:48 AM)
Thank you
From Lia Anggraini to Everyone: (11:48 AM)
thank you katharina
From Dedy Rahman Wijaya to Everyone: (11:48 AM)
thank you
From Georgios to Everyone: (11:51 AM)
Thank you
From Georgios to Everyone: (12:01 PM)
Good morning
From Tisa Siti Saadah to Everyone: (12:04 PM)
morning
From Sabrina to Everyone: (12:05 PM)
morning everyone
From FIF - Agung Toto Wibowo - 06810035 to Everyone: (12:08 PM)
morning all
From Operator CLOVE 10 to Everyone: (12:09 PM)
Feel free to ask on rocket chat at #practical5 channel
From Risnandar, Ph.D. to Everyone: (12:09 PM)
If any question or comment to this session, please chat in rocket
From Arif Qodari to Everyone: (12:25 PM)
got this error `ValueError: total size of new array must be unchanged, input_shape = [147], output_shape = [7, 7, 32]` could you go back up a bit please? decoder I think thanks, there was typo
From Risnandar, Ph.D. to Everyone: (12:27 PM)
yes, I think so typo
From Faldi Sulistiawan to Everyone: (12:33 PM)
what is the purpose of conv2dtranspose in the architecture? okay, thank you
From Georgios to Everyone: (12:37 PM)
In colour images, can we use variational Autoencoder to extract image colour information? Thank you
From Mawanda A to Everyone: (12:41 PM)
Can the distribution be other than Gaussian?
From Risnandar, Ph.D. to Everyone: (12:42 PM)
ok, next questions if any
From Mawanda A to Everyone: (12:44 PM)
What if the distribution that have more than 2 parameters? Like skewed distribution
From Faldi Sulistiawan to Everyone: (12:47 PM)
in all of the variation of VAE, do we just need the mean and logvar the input? or there is a kind of VAE that requires more than just that two parameters? okay, thank you
From Mawanda A to Everyone: (12:55 PM)
How can we compare the performance of the result from both of the models? Are there any specific metrics? AE and VAE Ok, I see
From Hariyanti Binti Mohd Saleh to Everyone: (12:58 PM)
Hi, boss Andit..how if we want to reduce KL loss weight...how to code..thanks..
From Arif Qodari to Everyone: (1:01 PM)
the plot_history seems has different scales than the actual loss value? in VAE
From Mawanda A to Everyone: (1:02 PM)
Sorry for slightly OOT, did you just mentioned about U-Net? How can we connect the layers to a model in keras if there is a skip connection in the layer?
From Georgios to Everyone: (1:04 PM)
Can you please show again the correction in Plot History?
From Mawanda A to Everyone: (1:08 PM)
Do we need to use low level or middle level of tensorflow instead of Keras API? For the skip connection things
From Georgios to Everyone: (1:11 PM)
Thank you
From Mawanda A to Everyone: (1:11 PM)
Thank you for the explanation
From Hariyanti Binti Mohd Saleh to Everyone: (1:12 PM)
Hi Boss Andit, is there any tips to be good in mathematics in Tensorflow? especially to play around with algorithm. any reference..thanks..
From Georgios to Everyone: (1:15 PM)
What is the GitHub link for the 5b? Thank you
From Risnandar, Ph.D. to Everyone: (1:15 PM)
adf-telkomuniv github
From Wawan Cenggoro to Everyone: (1:17 PM)
I found that it is easier to understand math in a paper by using Siraj's tips on how to read paper: https://www.youtube.com/watch?v=SHTOI0KtZnU
From Wawan Cenggoro to Everyone: (1:17 PM)
Though he is less credible now because of his scandal Another good video on how to read a research paper: https://www.youtube.com/watch?v=Uumd2zOOz60
From Risnandar, Ph.D. to Everyone: (1:20 PM)
thanks @wawan
From Hariyanti Binti Mohd Saleh to Everyone: (1:20 PM)
TQ
From Wawan Cenggoro to Everyone: (1:21 PM)
you're welcome
From Risnandar, Ph.D. to Everyone: (1:21 PM)
questions or comments or anything else, please feel free
From Hariyanti Binti Mohd Saleh to Everyone: (1:22 PM)
Btw, Wawan,, i don't have problem to understand in papers..but i'm facing problem when I want to implement the algorithm in TF.. any tips on that?
From Wawan Cenggoro to Everyone: (1:23 PM)
I see, sorry for the misunderstanding I find that tensorflow is somehow just an advanced version of numpy
From Wawan Cenggoro to Everyone: (1:24 PM)
I always code using tensorflow/pytorch by imagining the matrix manipulation exactly like if I code it using numpy
From Ebuka Oguchi to Everyone: (1:25 PM)
How do know the best loss function to use? How do we know how many times to train the generator per discriminator update
From Mawanda A to Everyone: (1:27 PM)
Are all the other variant of GANs based on the same Generator + Discriminator concept? Thank you, awesome
From Georgios to Everyone: (1:30 PM)
Thank you very much.
From Tuan Anh VU to Everyone: (1:30 PM)
thank you so much