⌨️ () Welcome to this course
⌨️ () Keras Course Introduction
⌨️ () Course Prerequisites
at when did we associate Labels with Image? we only created the images' batch but when did we associate with its labels?
I am sure I missed something, Still.... Where does that labels come from? ? How did it distinguish cats and dogs?
⌨️ () DEEPLIZARD Deep Learning Path
⌨️ () Course Resources
I finished off at
At , I think the model did a deep learning to see the espresso layer under the cappuccino top.
⌨️ () About Keras
⌨️ () Keras with TensorFlow - Data Processing for Neural Network Training
at as trial has 2100 participants, shouldn't the sample size be for i in range(50) and the other for i in range(2050) ? Also when running the 'model.fit', I am not getting the 2100/2100 as shown in 28:35. Instead getting 378/378. Hopefully somebody can help me see clearer. Thank you
the for loops are not clear to me. The for loops have been divided (1-50) and (1-1000). 5% of 2100 = 105, and 95% of 2100 = 1995. What is the logic behind dividing the for loops in such a manner?
it should be "for i in range(950)" NOT "for i in range(1000)"Reason: if 5% = 50 person, then 1% = 5 person. So, 95% = 950 person.
why were the ranges in the first and second for loop different ? First was 50, second was 1000. Why ?
Hi, why to rescale the numpy array, I didnt understood well? ()
my results returned either 0. or 1. but not a decimal number between 0 and 1?
⌨️ () Create an Artificial Neural Network with TensorFlow's Keras API
Also, is Ultra CUTE <3
can some one explain me about () input_shape in :Dense(units=16, input_shape=(1,),activation='relu')
⌨️ () Train an Artificial Neural Network with TensorFlow's Keras API
, the parameter "learning_rate" has been renamed to "lr" from version 2.2.* to 2.3.0 in September 2018
I’m not seeing the same output - it only prints out one epoch and then fails ?
Hi Mandy, I tried training neural network for the clinical trials example @. I am getting Nan for loss and 0.5 for accuracy. I ran the exact lines of code you did. What is the issue? BTW, I am not using the three lines of code to use the GPU.
Hi, I'm new to the subject and eager to understand. ☺️ Watching the episode Train an Artificial Neural Network (circa ) and I understand the training but don't really understand whether the model there was just trained "theoretically", so to say, or was there any actual data input present, as I don't remember that being discussed. I understand that information is being added gradually and I can live with not knowing everything at once 😄, but this seems important - can I run a model without any actual input to check how it works or is it just not being shown. Sorry if I'm misunderstanding everything 😄😅
⌨️ () Build a Validation Set With TensorFlow's Keras API
⌨️ () Neural Network Predictions with TensorFlow's Keras API
At if the test set should be scaled the same way the train set was, shouldn't we be using scaler.transform (i.e. without fitting the scaler again).
you shouldn't call fit_transform on test_samples in . You should use the same scaler that was fitted on train_samples.
Thank you for this nice video, it really got me started. But I think there is one bad habit involved: "Never normalize your test data on its own but rather on the train data!" (This is what you read in many forums and for example in Chollet's "Deep Learning with Python" book. I think at it should be scaler.transform(test_samples.reshape(-1,1)) as this takes the fitted scaler from the train_samples. Correct me if I am wrong :)
why need batch_size for prediction?
⌨️ () Create a Confusion Matrix for Neural Network Predictions
@ lol 420 woohoo
⌨️ () Save and Load a Model with TensorFlow's Keras API
too cute
⌨️ () Image Preparation for CNNs with TensorFlow's Keras API
I think all the photos were meant to be directly under the dogs-vs-cats folder as per the later demonstration. Cause at 1:08:33, all the remaining photos were directly under dogs-vs-cats folder. Not dogs-vs-cats/train.
/ 2:47:54
did u get past the for loops part? at
When she did "flow_from_directory" at she passed in a list of labels there which are then saved in the "train_batches" Generator object. When she does "next(train_batches)" she gets access to the actual images and labels in the batch. She talks a bit more about this at 1:26:00
Shuffling in Test dataset VGG: At , Mandy says not to shuffle test dataset so that confusion matrix remains valid. I didn't get this, why would shuffling test dataset cause any problem??
what has shuffling to do with confusion matrix?
how did we obtain the labels for the images ?
One simple question @, how to know that 1 represents a dog and 0 represents a cat. Is it by comparing the pictures with the labels from the output of the plotImages function or is there some way to know it
⌨️ () Build and Train a CNN with TensorFlow's Keras API
@ you mentioned padding='same' means images will have 0-padding. That's incorrect, right? Same padding means that the output has the same size as input. You do mention that. Just that it won't be 0-padding. p = (k-1)/2.
@ I am getting error Error "when checking input: expected conv2d_3_input to have shape (224, 224, 3) but got array with shape (244, 244, 3)" How to resolve this?
⌨️ () CNN Predictions with TensorFlow's Keras API
- I don't get what she means. Maybe it didn't come out well. She says we don't want to shuffle the test set to plot the confusion matrix but from what she says it seems that the problem is that we don't have a 1-to-1 match between labels and samples which is not true. So why does the confusion matrix requires the set to be unshuffled?
⌨️ () Build a Fine-Tuned Neural Network with TensorFlow's Keras API
I am getting error in in [83] run in for loop.it says :Input 0 of layer block1_conv1 is incompatible with the layer: : expected min_ndim=4, found ndim=2. Full shape received: [None, 2]
⌨️ () Train a Fine-Tuned Neural Network with TensorFlow's Keras API
to the model in (vgg16)one run per epoch needs nearly 5 min for me while it takes just a few seconds for you. do you know why? (maybe i do not use a gpu?)
⌨️ () Predict with a Fine-Tuned Neural Network with TensorFlow's Keras API
⌨️ () MobileNet Image Classification with TensorFlow's Keras API
... classifying lizard 🦎
I though it was called deeplizard because she looks like a lizard (also though that was the reason the lizard in the logo was female), but I guess that was just an indirect reason.
- Expresso is just the coffee. It becomes a cappuccino when you add milk and foam. If it's getting classified as expresso then I wonder if the original dataset labels were added incorrectly by human editors that didn't know the difference.
⌨️ () Process Images for Fine-Tuned MobileNet with TensorFlow's Keras API
⌨️ () Fine-Tuning MobileNet on Custom Data Set with TensorFlow's Keras API
You need to add one more layer before the Dense output with 10 neurons since the chosen modified layer is not working as for today (22nd July 2022). You can add it as x = tf.keras.layers.Flatten()(x) . Hope this helps.
- When I run the same code I get different numbers. Total params is the same but I get:2,136,074 Trainable params1,103,040 Non Trainable params.
At ❤️ cute
⌨️ () Data Augmentation with TensorFlow' Keras API
Nice.
your cam got augmented too (and flipped upwards) :DNonetheless awesome course!
