Deep Learning mit Linux

So I want to have a simplistic but fully featured ML setup
my specs:

• needed python version
• conda instead of pip (as long as possible)
• linux os
• tensorflow
• keras
• jupyter notebooks
• pycharm IDE

I am following
https://www.pyimagesearch.com/2019/12/09/how-to-install-tensorflow-2-0-on-ubuntu/
(to lengthy / decided not to use gpu for teaching myself)
instead I used straightforward:

• https://www.tensorflow.org/install/
• tensorflow hello world (didn't get it running!)
•  https://github.com/aymericdamien/TensorFlow-Examples/blob/master/examples/1_Introduction/helloworld.py
• following error apeared, couln't solve: 
• sess = tf.Session()
  Traceback (most recent call last):
    File "<stdin>", line 1, in <module>
  AttributeError: module 'tensorflow' has no attribute 'Session'
• cpu tensorflow/stream_executor/cuda/cuda_driver.cc:351] failed call to cuInit: UNKNOWN ERROR (303)
• https://keras.io/getting-started/sequential-model-guide/
• keras- hello world: 
• https://gist.github.com/Orbifold/3f51f3c21d348873f4025813bfcf5614

confused me:

• not to use python2.7 but  python3.7
• to use pip instead of pip2

install Pycharm best by Ubuntu Softwarestore

1. install python3
2. pip install virtualenv # install virtual environments
3. #for each projet: create environment
• mkdir project ; cd project
• python3 -m venv venv
• source venv/bin/activate
4. pip install tensorflow
5. pip install keras
6. pip install ipykernel
7. python -m ipykernel install --user --name=venv # install ipykernel
   

in Jupyter change kernel to virtual envirionment click:
Kernel->chang kernel->venv

Keras has become a part of tensorflow. The book's code must be adapted:

replace

• import keras

by

• import tensorflow.keras as keras

Everything then should work fine.

Otherwise, you can replace each code of keras:

Access keras functionality through tensorflow:
instead of

• import keras
• model = keras.Sequential()

do

• import tensorflow as tf
• model = tf.keras.Sequential()

With these instructions everything should work out of the box


To start jupyter:

• jupyter notebook

To leave environment
 deactivate

a good