Sie sind auf Seite 1von 1

I begin my expereince of machine learning with supervised machine learning.

I came through linear regression, logistic regression and most interesting model
, Neural Network.
I used Octave as my developing tool and developed several ML algo on it.
I solved the classification problem of detecting the disease as chronic or non c
hronic using sigmoid function,
having 5 features.
Next, I have implemented the same with Neural Network using one hidden layer and
one output layer with 2 units(0 and 1).
Training the parameters efficiently is the most important thing.
I used advanced optimization function to train the parameters. I also tried usin
g Normal equation to train the parameters.
I studied several mathematical derivation - Bernoulli's theorem, Gaussian distri
bution of data.
I have also implemented Kernel algo for depicting mail as spam or not.
I came through the problem of high biassed and high varinace and solved it by u
sing regularization parameter.

Das könnte Ihnen auch gefallen