- Write a function to produce the convolution of an image with a filter - Write a function that fills a buffer with a rolling set of image object segmentation results - From a set of loss functions like L2 loss or hinge loss, explain which ones are rotationally invariant, sensitive to rescalings of the input features, and which are convex. - Given a set of standard random forest trees and a set of gradient boosted trees, explain how the model's predictions would change if you remove the first tree, the last tree, or make other changes after training. - Talk about a system design you've worked on where a web service provides computer vision computations on input images. How did you test it? How did you deploy it? - Given a pandas DataFrame with some numeric columns, how to produce counts of data for different category IDs, including with or without extra processing to remove noisy, missing, or outlier data points.
