How to initialize adaboost's weighted distribution?

Question

I have a homework assignment about implementing the adaboost algorithm.
We are following this article by Schapire. I'm not exactly sure what they mean in the assignment by :

Let the training patterns and their labels in dataset D be denoted xi and yi, respectively and let Wk(i) be the kth (discrete) distribution over all these training samples. We initialize these weights Wk to to be uniform in iteration k=1.

Do I just make an array of ones?

Wk = ones((size(H(x)))

Then updated it accordingly on each iteration.

Answer 1

Discrete uniform distribution means your total probability mass, which it is equal with $1$ by definition, is split in equal quantities for each event. For your case, the events are the observations. Thus, if you denote by $N$ the number of observations, you will have $W_{k(i)}=\frac{1}{N}$.