Suppose we performed a single loop and updated the weights for some synthetic dataset:
| Sample | The updated weight |
|---|---|
| Sample 1 | 0.08 |
| Sample 2 | 0.34 |
| Sample 3 | 0.08 |
| Sample 4 | 0.08 |
| Sample 5 | 0.34 |
| Sample 6 | 0.08 |
To construct the new dataset, the following 6 random numbers were chosen: [0.16, 0.59, 0.65, 0.01, 0.35, 0.93]. Build the buckets based on the updated weights and use the 6 generated floats to construct the new dataset for the next iteration of the AdaBoost algorithm (look up the optional step in the walkthrough example). The answer should contain the indexes of the samples that were chosen more than once, separated by a space.
For example, if sample 3 and sample 4 were selected twice, the answer would look like this: 3 4