Machine Learning for OpenCV

Fitting the perceptron to data

We can instantiate our perceptron object similar to other classifiers we encountered with OpenCV:

In [6]: p = Perceptron(lr=0.1, n_iter=10)

Here, we chose a learning rate of 0.1 and told the perceptron to terminate after 10 iterations. These values are chosen rather arbitrarily at this point, although we will come back to them in a little while.

Choosing an appropriate learning rate is critical, but it's not always clear what the most appropriate choice is. The learning rate determines how fast or slow we move towards the optimal weight coefficients. If the learning rate is too large, we might accidentally skip the optimal solution. If it is too small, we will need a large number of iterations to converge ...