Inverted Dropout (Seeded)

Problem

Inverted dropout in training mode: call random.seed(seed), then keep each element with probability 1 - p and scale survivors by 1 / (1 - p); dropped elements become 0.0.

The signature is inverted_dropout(x, p, seed). Iterate the elements in order so the output is reproducible.

Hint

Seed first, draw one random number per element, and scale kept values to preserve expectation.

Test cases

case 1

in: [[1,2,3,4,5],0.5,0]

expected: [2,4,0,0,10]

case 2

in: [[1,2,3],0,42]

expected: [1,2,3]

case 3

in: [[0.5,1.5,2.5],0.25,7]

expected: [0.6666666667,0,3.3333333333]

case 4

in: [[-1,2],0.8,123]

expected: [0,0]

Starter codePython

import random
def inverted_dropout(x, p, seed):
    # Your code here
    pass

4 test cases · run your code against them in the editor