Formulas

Question 1

What is the formula for mean squared error?

Answer 1

Question 2

What is the sigmoid function?

Answer 2

Question 3

What is the function for the hyperbolic tangent?

Answer 3

Question 4

What are the two equations used to update the weights using Momentum?

Answer 4

Question 5

What is the equation for the running average of the gradients used in Adam?

Answer 5

Question 6

What is the equation for the squared gradients used in Adam?

Answer 6

Question 7

How is each parameter updated when using Adam?

Answer 7

Question 8

What is the Bayes Rule?

Answer 8

Question 9

What is the formula for the entropy of a discrete probability distribution?

Answer 9

Question 10

What is the formula for KL-divergence for two probability distributions?

Answer 10

Question 11

What is the formula for the entropy of a continuous probability distribution?

Answer 11

Question 12

What is the formula for the KL-divergence of a continuous probability distribution?

Answer 12

Question 13

What is the entropy of a Gaussian Distribution?

Answer 13

Question 14

What is the entropy of a d-dimensional Gaussian distribution?

Answer 14

Question 15

What is the KL-divergence between two d-dimensional multivariate Gaussian Distributions?

Answer 15

Question 16

What is the Wasserstein difference for two multivariate Gaussian Distributions?

Answer 16

Question 17

What is the cross entropy error for a binary classification task?

Answer 17

Question 18

What is the Gaussian Distribution equation?

Answer 18

Question 19

What is the multivariate Gaussian Distribution equation?

Answer 19

Question 20

For softmax, what is Prob(i)?

Answer 20

Question 21

For softmax what is log Prob(i)

Answer 21

Question 22

What is the equation for the gradient using softmax?

Answer 22

Question 23

What is the weight decay equation?

Answer 23

Question 24

What is the formula for the mean?

Answer 24

Question 25

What is the formula for the variance?

Answer 25

Question 26

What is the equation used for batch normalisation?

Answer 26

Question 27

What is the loss function for Neural Style Transfer?

Answer 27

Question 28

What is the true value (V*) of the current state?

Answer 28

Question 29

What is the formula for Q*(s,a)?

Answer 29

Question 30

What is the formula for the fitness of a policy?

Answer 30

Question 31

What are we trying to minimise with Double Q-learning?

Answer 31

Question 32

What is the formula for the Advantage function?

Answer 32

Question 33

What is the equation for GANs?

Answer 33

Question 34

What is the threshold activation function (context of this course)

Answer 34

Greater than the bias = 1, less than the bias = 0

Question 35

In a deterministic environment, with a learning rate of 1, what is the Q-learning update rule?

Answer 35

Question 36

Write the formula for activation Z of the node at location (j, k) in the ith filter of a Convolutional Neural Network which is connected by weights K to all nodes in a M x N window from the L filters (or channels) in the previous layer, assuming the bias weights are included in the activation function g().

Answer 36

Question 37

What is the formula for the number of free parameters in a CNN layer?

Answer 37

F x (1 + L x M x N ), where F is the number of filters, M x N is the filter size, and L is the number of channels

Question 38

What are the functions that describe a LSTM?

Answer 38

Question 39

What is the Variational Auto Encoder trained to maximise?

Answer 39

Question 40

For GANs, what is the formula for V(G, D)?

Answer 40

Question 41

What is the formula for number of weights per filter?

Answer 41

(filter width) x (filter height) x (input depth) + 1 (for bias)

Question 42

Number of neurons in this layer?

Answer 42

(output width) x (output height) x (depth)

Question 43

Number of connections into the neuron’s in a layer?

Answer 43

(num neurons) x (connections per neurons / filter wights - bias)

Question 44

Number of independent parameters?

Answer 44

(num filters) x (num weights)