Part 05 - Constructing a Neural Network Models - Regression model with Python (Tensorflow & Keras)

Regression models.
1. Construct a multi-layer neural network model.
2. Tune the hyper-parameters for training a model.
3. Train a model on CPU or GPU.
4. Evaluate a model.
5. Save and load a model.
Notes:
All of the code in this tutorial implemented in Tensorflow 2.1.0.
The source code is available here:
github.com/zha...
Zeyad Hailat

[@karapureddyjaswanthreddy6053]

This is one of the best lectures on this topic. Great work.

[@prathapkb123]

Wow... The best hands-on explanation in youtube so far. Good work bro. Thanks a lot. Learnt a lot!!

[@prywatny]

Excellent work! Thanks!

[@melissa1343]

thank you, im wondering if we can get the code for this in github or something ?

[@MOUADKARMOUN-q2p]

hello i have a problem when i run the code with same data that you give, i got NaN value for loss mae mse mape val_loss val_mae val_mse val_mape

[@sameutbouhaikizzeddine8238]

I have the same problem!

[@mitajanaiitm]

I am also getting same problem of NaN

[@leventoz1408]

Very nice!!!

[@data_scientists_learning]

Thanks!!

[@vedikaadhiman388]

Thank you for this explanatory tutorial. I have also followed the same code for building ANN model on regression data having 4 input variables and 1 output variable and the model is built successfully. But now, I want to optimize the ANN model's solutions by Genetic Algorithm (GA). So, for this I am following "kzbin.info/www/bejne/opvNl6WofdGredksi=-DBs2Bki0p1LgABm", this tutorial as a reference but, they have used Random Classifier in place of ANN model to optimize their regression data. So, at their 19th step, they are calling their trained model into a new variable which is used later as a "fitness function" for GA. I am using "model_full = ann_viz(model, view=False, filename= 'network.gv', title='My Neural Network')" instead of their Rnadomforest classifier. So, please help me in this as I am getting error in "input_shape" i.e., "AttributeError: 'Dense' object has no attribute 'input_shape'". Although I have changed the input_shape as mentioned by you while building ANN modelto "4" as I have 4 variables.

[@mitajanaiitm]

while running the code I am getting all loss values as NaN, filepath I have mentioned as suggested in runtime checkpoint_path = '/tmp/ckpt/checkpoint.model.keras.weights.h5'