Where is the next part of this video

#1

this video is incomplete as till now we have made validation set .where is the next part where we actually train our data?

#2

hey @Par1hsharma ,
i am not sure how your course structure is showing it, but it should be the next lecture itself.

But stil, try once from your side to train it and then let me know if there are some issues / doubts on it.

#3

https://drive.google.com/file/d/1otJygXVDE6CEUKvvQZbmkSNEWLDi3fcL/view?usp=sharing
please see this I dont have next video of training this.the next part is a YouTube video of imgaug library

#4

you would have been taught about tranining in other videos also …no ?

#5

yes but I dont know what layers should I add and how to make structure of cnn model for this dataset . training was for other datasets

#6

So , About Adding layers and all. No body can teach you that , its just an experience and experiments.
With working on different models , you will get to know how they have used different layers one after the other , there pattern and all.
and before that , you need to learn what each layer do and how to be used.

Once , you get them , then you can go ahead and start making models.

#7

well there must be a approach which needs to be followed to build models ?

#8

no @Par1hsharma .

There is never any general approach.

Based on the data , the understandings, you yourself decide which layer to try on.
No one can tell you that at this point you should use this layer .

But yeah , one thing is …
if you want to learn more features then increase the number of nodes in that layer.
so, you data is of 200 features , then i will use a dense layers of 512 first , and then lower to perform my task.
similarly in CNN, filters are changed as 32-64-128-512 …---- to learn information.