3-layer Neural Network means a total of 3 layers including input and output layer. However, could you please tell me which part of the lecture you are referring to?
Its fascinating how mechanics's momentum, moments etc concepts are adopted in ML & Deep Learning. Sandeep sir is awsome the way he drills down every layer of complex topics. I am gonna through this whole course again after having my certificate.
Make sure you're using a internet connection with good download and upload speeds.
Also please avoid opening notebooks in multiple tabs.
If you're still having issues please send a screenshot at reachus@cloudxlab.com
Any ML/DL model is basically an approximation of what the original value is. So, when it is predicting the training set with 100% accuracy, that means it is memorizing everything, or in other words it is overfitting.
Somewhere it is written bool training="training" , somewhere training="true", and traning="false".
I have understood training = "true" we use while we are training and "false" when we are run for validation
When building a machine learning model it is often convenient to distinguish between variables holding trainable model parameters and other variables such as a step variable used to count training steps. To make this easier, the variable constructor supports a trainable=<bool> parameter.
The dropout is applied on the layer of neurons. Dropout is essentially turning off some of the neurons. So, you can say, normalization comes before dropout.
Please login to comment
23 Comments
Sir, I need complete code of freezing the layers from scratch
Upvote ShareHi,
Could you please elaborate a little more on the issue you are facing?
Thanks.
Upvote Sharehi,
When you say "three layer neural network", do you mean 3 hidden layers or 1 input and two hidden layer?
Since in dropout example, while implementing dropout regularization to 3 layer neural network. we have created only 2 hidden layers.
while in previos discussions in introduction to ANN we were having 2 hidden layers for two layer neural network.
Upvote ShareHi,
3-layer Neural Network means a total of 3 layers including input and output layer. However, could you please tell me which part of the lecture you are referring to?
Thanks.
Upvote ShareIts fascinating how mechanics's momentum, moments etc concepts are adopted in ML & Deep Learning. Sandeep sir is awsome the way he drills down every layer of complex topics. I am gonna through this whole course again after having my certificate.
Regards,
HS
1 Upvote ShareThis comment has been removed.
The jupyter lab is continuosly getting disconnected could you please do something for this??
Upvote ShareHi Sharathchandran,
Make sure you're using a internet connection with good download and upload speeds.
Upvote ShareAlso please avoid opening notebooks in multiple tabs.
If you're still having issues please send a screenshot at reachus@cloudxlab.com
I want to know that when Batch accuracy comes out to be 1.0, does this means overfitting?
Upvote ShareHi,
Any ML/DL model is basically an approximation of what the original value is. So, when it is predicting the training set with 100% accuracy, that means it is memorizing everything, or in other words it is overfitting.
Thanks.
Upvote ShareSomewhere it is written bool training="training" , somewhere training="true", and traning="false".
I have understood training = "true" we use while we are training and "false" when we are run for validation
But when do we use training="training"
Upvote ShareHi,
It is referring to a boolean value.
Thanks.
Upvote Sharewhat does this refer???? training="training" true or false? or we are just initialising it
Upvote ShareHi,
Could you please tell me where is it referring to training= and not trainbale=?
Thanks.
Upvote ShareHi,
global_step=tf.variable(0,trainable=False)
at the video position 1:1:33 what does this trainable=False means?
Upvote ShareHi,
When building a machine learning model it is often convenient to distinguish between variables holding trainable model parameters and other variables such as a
stepvariable used to count training steps. To make this easier, the variable constructor supports atrainable=<bool>parameter.Thanks.
Upvote ShareThis comment has been removed.
Hi,
So book can take either values; true or false. Also, it is to be noted that A = B is not the same as A == B.
Thanks.
Upvote ShareHi,
In AdaGrad we do element wise calculation, but in Gradient descent,Momentum opt.,nesterov accelerated gradient do we also do element wise calculation?
Upvote ShareHi,
Here is an elaborate explanation of the differences:
https://stackoverflow.com/questions/36162180/gradient-descent-vs-adagrad-vs-momentum-in-tensorflow
Thanks.
Upvote ShareWhat is the correct order of applying batch norm and drop out? Which layer comes first, batch norm or drop out?
Upvote ShareGood Question.
The flow is like this:
Input -> Normalization -> Neurons Layer -> Output -> Batch Normalization -> Another Neurons Layer
The dropout is applied on the layer of neurons. Dropout is essentially turning off some of the neurons. So, you can say, normalization comes before dropout.
Upvote Share