The next block of codemy tutorials sets up an actual machine learning
model, using pytorch. This model will be using the - by now quite famous -
Iris data set. New to me, the dataset goes back to 1936, when first efforts
were being made to quantify species identification in botany. The
readings on two of the specieswere done in the Gaspésie region of Quebec,
all on the same day...
The set consists of fifty samples each of three species of Iris. By measuring
sepal height and with, and petal height and width. There is some overlap, but
one could discriminate between each species after doing some math. In Machine
Learning, the computer does all the work.
From kaggle:
* * *
The model will be using Relu activation , which has become standard with
more powerful computers.(The activation function decides what data will be pushed
forward). Relu is zero-centered and negative readings are all set at zero.
* * *
Codemy's pytorch5 defines the model, using OOP and thus a class.
No comments:
Post a Comment