Neural Network Predictions with TensorFlow's Keras API

[@deeplizard]

👉 Check out the blog post and other resources for this video: 🔗 deeplizard.com/learn/video/2f-NjDUvZIE 👀 Come say hey to us on OUR VLOG: 🔗 kzbin.info

[@syedrehman5755]

huge respect from Pakistan for such a great teaching with beautiful voice and accent

[@tymothylim6550]

Thank you very much for this video! It was a great step-by-step video to show the implementation of making predictions (in this case with the validation set)! It's a good buildup from the previous video!

[@mahmoudabdulhamid4182]

we should use transform instead of fit transform scaled_test_samples = scaler.transform(test_samples.reshape(-1,1))

[@Sikuq]

Great video. And thanks to your Patreon member help, I am now running this source code beautifully in VS Code. Love it.

[@deeplizard]

Thanks, Christian!

[@rakeshsinghrawat99]

Thank you

[@juanodonnell]

In [23] shouldnt you be using transform instead of fit_transform?

[@nikitarazguliaev1715]

Yes, though in this particular example it doesn't matter, as ranges of features are the same in train and test dataset

[@yepnah3514]

hi your videos are helping me with my school project! i need to create a test for my horse-human model. im confused on where to start watching. which of your videos i need to start watching so i can create my test? thanks a lot!!

[@fernandojackson7207]

Excelent. Is there a way of numbering the prediction outputs so that we know, e.g., the 37th data point is a 1 or 0?

[@vishalkuber3631]

hey.. how can i create confusion matrix for stock data..?

[@nikitabisht2605]

Why have we used 10 and 200 for loops?

[@GauravSingh-ku5xy]

Why did we shuffle test samples and labels during the predictions?

[@aswink4300]

You guys are explaining very well, can you add up some illustrated explanation like your initial videos

[@deeplizard]

Thank you, Aswin! Illustrations don't have as much of a place in this course since we're focusing mostly on the code and not on the underlying concepts. Our Deep Learning Fundamentals course is full of illustrations to explain the concepts :D deeplizard.com/learn/playlist/PLZbbT5o_s2xq7LwI2y8_QtvuXZedL6tQU

[@aswink4300]

@@deeplizard thanks so much

[@lividpudding8565]

Could you please share the name of the music that plays at the end?

[@eapresents6061]

Thanks for great explanation. Well done! After the theoretical parts are concluded, are you planning to work on real world applications and sharing them?

[@deeplizard]

Yes, we will work on basic real world applications in the course. More advanced applications will likely be covered in separate courses since this course is aimed at beginners :)

[@sendjasniabderrezzaq9347]

Thanks for the explanation, it is really intuitive. The results I got in my machine are quite different from yours and it doesn't make any sense to me. The loss and accuracy value between the first epoch and the last one is not much different. Epoch 1/30 loss: 0.7036 - accuracy: 0.4963 Epoch 30/30 loss: 0.7023 - accuracy: 0.4974 What could be the reason?

[@deeplizard]

I would assume something is not matching between my code and yours. Perhaps you generated the data differently or missed a line of code processing the data somewhere?

[@sendjasniabderrezzaq9347]

@@deeplizard Thanks for your response. I didn't have the time to check the code. Everything worked perfectly for the rest of the tutorials, so I assumed that something were wrong like you said. BTW I followed all episode on Keras, It was very helpful, thanks for that.

[@harchitb]

1. How is validating different from testing 2. What is happening in the math when predicting ?

[@jay-rathod-01]

Hello ma'am if possible could you please make a video also on SVM other things except NN in ML?

[@deeplizard]

Perhaps in the future :)

[@ineshtickoo2530]

Amazing video! I noticed one thing tho, when we're talking about 5% of the people being 10 people, that'd mean that the TOTAL number of people would be 200, hence the 95% sect would be 190 instead of 200. just an observation hahah. very clearly explained tho 10/10

[@andreyelven8095]

Hello! I got a question. How can we have: train_labels, train_samples = shuffle(train_labels, train_samples)? It's kinda weird for me, because we have two variables and only one function which returns something. How is it even possible? I tried to have something like a, b = 1, but it's an error. I totally understand what this function does, so the question is about syntax.

[@mdyounusahamed6668]

what is the difference between validation set and test set?

[@deeplizard]

deeplizard.com/learn/video/Zi-0rlM4RDs

[@mdyounusahamed6668]

@@deeplizard Thank you. You are very helpful.