Welcome

Let’s go!

Stuff to do.

• Get started with the bachelor thesis

• Look up line search.

• Try the housing dataset with a forward/backward AIC/BIC algorithm. Then run the remaning dataset thru a tensorflow model.

• Finish the munta latex.
  

Questions:

Math

• What statistical meaning does the different activation function have in a neural network?

Programming

• How can you use Obj-c to get information for a flask/Django app?

This list last updated 22/7 04:41

In progress

• Should we use $x, x^{2}$ and so when doing data analysis?

  Yes check out statistical “interaction” in multivariable linear regression.

• What is the optimal way of calculating weights is there another way then gradient decent?

A neural network with a lot of deep layers becomes more of a “approximation” then a “statistical regression”. But to increase the possibility of a steeper decline we can take the variables that have the most correlation and is not correlated to any other. Then in we can check that if we add a $x^2$ term to increase the explained variance and decrease the variance of the error. (Check out the anova variance stuff.)

Done

• How do you connect a (Flask)/Django app to postgres or another SQL-service? It’s is built into Django. Google models Django.