3. Suppose a simple single layer perceptron with 2-4 input X=(x1, x2), a bias, ReLU for activation function, and MSE (mean squared error) for loss function. DDDD X 2130 X 1 2 0 -1 a. Show evolution of weights according to stochastic gradients algorithm with batch size = 1 b. Show evolution of weights according to stochastic gradients algorithm with batch size=2