High Level Vision

Question 1

Describe the steps for image classification:

Answer 1

1) In training stage, pass labelled images through classifier to extract features
2) the model produces a prediction for each category
3) calculate the loss between the predictions and the ground truth
4) backpropagate and update the parameters accordingly
5) after many iterations the model will converge
6) in the test stage, the model is fixed. Pass the test image through the model to get the prediction

Question 2

What is binary classification?

Answer 2

Classifying if an image is a specific object or not. Results in “tiger” and “non tiger”

Question 3

What is multiclass classification?

Answer 3

Dataset contains multiple categories. Given an image, the classifier assigns a label to it.

Question 4

What is multi-label classification?

Answer 4

Images contain multiple objects, aim to predict the probability for all the objects the image contains

Question 5

What is hierarchical image classification?

Answer 5

First the classifier predicts the wider category the image contains, then it tries to label it more specifically with subcategories
e.g. Fruit, apple, ladywell apple

Question 6

What is a basic approach to image classification?

Answer 6

given RGB image with 32323 pixels, 10 categories, predict 10 numbers representing probability of category in image.
- Summation of probabilities = 1

Question 7

What does f(x, W) equal?

Answer 7

f(x, W) = (W*x) + b
where x is the image
W is the weights/parameters
b is the bias

Question 8

How would we get the predictions for this: given RGB image with 32323 pixels, 10 categories

Answer 8

f(x, W) = (103027 vector)(13027 vector) + (10 1 vector)

Question 9

How would you calculate the score for an image with 4 pixels and 3 classes:

Answer 9

1) Flatten image into 1D vector (4x1)
2) perform matrix multiplication with weights for each category (3x4)
((3x4) . (4x1) = (3x1)
3) add bias to (3x1) vector to get final prediction for each category

Question 10

How does the process above change if you use multi-layer perception?

Answer 10

For each different category the weights and bias is different. So for 3 categories you would multiply the (4x1) vector with a different (1x4) grid of weights, then add the bias

Question 11

How do we find good values for W and b?

Answer 11

• Start with random values then converge to the optimal values of W and b that minimise loss

Question 12

What is a loss function and what methods could we use to calculate loss?

Answer 12

• A loss function tells us how accurate the classifier is at predicting the categories.
• Large loss indicates a poorly trained classifier

Could use the L1 or L2 loss, or SVM loss, Cross-entropy loss, MSE loss, Softmax loss.

Question 13

What would be the formula for calculating L1 loss:

Answer 13

• Calculating the loss over the dataset as the average of loss over images.
• 1/N* sum of loss for predictions of each individual image

Question 14

How is SVM loss calculated?

Answer 14

• takes the max between 0 and the value of the score of the non-actual label minus the score of the actual label + a margin delta (usually 1)

Question 15

If this is the prediction for values for 3 classes are given in the following table. Compute the multiclass SVM loss for each class. Then compute the total loss for all classes. Delta = 1

cat: 3.1 1.5 5.2
dog: 0.7 2.4 1.2
person: 1.5 5.1 -1.4

Answer 15

max(0, 0.7 - 3.1 + 1) + max(0, 1.5 - 3.1 + 1)= 0 + 0 = 0
max(0, 1.5 - 2.4 + 1) + max(0, 5.1 - 2.4 + 1)= 0.1 + 3.7= 3.8
max(0, 5.2- -1.4+ 1) + max(0, 1.2- -1.4 + 1)= 7.6+3.6= 11.2

Average: (0 + 3.8 + 11.2)/3 = 5

Question 16

What is the difference between a deep learning neural network and a simple neural network?

Answer 16

In deep learning there may be hundreds of hidden layers, that are used to train the model and produce the output

Question 17

What is a convolutional neural network?

Answer 17

• a type of deep learning model that uses convolutional layers that apply filters to input data to capture image features
• given an image and a filter it calculates the output, used for image classification

Question 18

What are recurrent networks?

Answer 18

The output becomes the next input.
- they have connections that form directed cycles.
- allows retention of memory of previous inputs through hidden states

Question 19

How do artificial neural networks work?

Answer 19

• neurons receive multiple inputs, which have adjustable weights.
• a threshold decides whether or not a neuron is active or not.

Question 20

What is the input signal formula for a neuron?

Answer 20

(sum of (weights*inputs)) + bias

Question 21

What is the output signal formula for a neuron?

Answer 21

y = function(input)

Question 22

What is an activation function?

Describe the threshold activation function:

Answer 22

It determines if a neuron is active or not

• choose a threshold, if the weighted sum of inputs + bias meets the threshold, the neuron is active

Question 23

What is the sigmoid function?

Answer 23

• It’s an activation function
• formula: x = 1/1+e^-x
• output is always between 0 and 1

Question 24

What are the two parameters that can be introduced to the sigmoid function?

How does it fit into the formula?

Answer 24

c1: controls slope of sigmoid
c2: controls horizontal offset

1/1+e^-c1*(x-c2)

Question 25

How do different values of x effect the output of the sigmoid function?

Answer 25

- for large negative inputs (x), the output approaches 0 - for large positive inputs (x), the output approaches 1. - for inputs near 0, the output is around 0.5

Question 26

What are the properties of the sigmoid function?

Answer 26

it's domain is (−∞, ∞) it's range is between and including (0, 1) when input is 0, output is 1

Question 27

What is the derivative of the sigmoid function?

Answer 27

σ′(x)=σ(x)(1−σ(x)) - include derivation process on 2D paper

Question 28

Name some other activation functions:

Answer 28

- Leaky Relu: max(0.1x, x) - tanh(x) - Relu max(0, x)

Question 29

What are TLUs?

Answer 29

Technical logic units - simplified version of threshold neuron model - they only accept binary inputs and the weights are usually = 1

Question 30

Describe the AND TLU node:

Answer 30

- 2 inputs are either 0 or 1, weights = 1. - range of outputs is [0,2] - threshold has to be a value > 1 [1,2] - if input*weight = 0, 0, output = 0 (not active as < threshold) - if input*weight = 0, 1 or 1,0, output = 0 (not active as < threshold) - if input*weight = 1, 1, output = 1 (active as >= threshold)

Question 31

Describe the OR TLU node:

Answer 31

- 2 inputs are either 0 or 1, weights = 1. - range of outputs is [0,2] - threshold has to be a value >= 1 [0,1] - if input*weight = 0, 0, output = 0 (not active as < threshold) - if input*weight = 0, 1 or 1,0, output = 1 (active as >= threshold) - if input*weight = 1, 1, output = 1 (active as >= threshold)

Question 32

Why is it impossible to have a single neuron with an XOR activation function?

Answer 32

Because XOR requires a restricted output region, or it would need two different thresholds. output would have be 0 < 1 < 2

Question 33

How can we create an activation function with a TLU of XOR logic?

Answer 33

Combine multiple TLUs - XOR Logic = (OR) AND (NAND) - 3 TLUs combined OR: 0 or 0 = 0 0 or 1 = 1 1 or 0 = 1 1 or 1 = 1 NAND: 0 and 0 = not(0) = 1 0 and 1 = not(0) = 1 1 and 0 = not(0) = 1 1 and 1 = not(1) = 0 OR AND NAND: 0 and 1 = 0 1 and 1 = 1 1 and 1 = 1 1 and 0 = 0

Question 34

What does a model generalising well mean?

Answer 34

That the model weights/parameters doesn't become too specific at knowing the features of the training data and is able to perform well on new unseen images as well.

Question 35

What happens if we use a function with too small a degree, in the middle and a very high degree?

Answer 35

- linear line (degree 1), small margin between data points - polynomial line (degree 2), we get accuracy and gradient stability - polynomial line (degree 9), we get accuracy but gradient instability (not a good option)

Question 36

What can happen if you train a model on training data too much?

Answer 36

You can achieve 99% accuracy/1% loss, but the model will be overfitted to the training data and not generalise/perform well on the test data

Question 37

What is the top-5 metric?

Answer 37

When calculating accuracy using correctly labelled image/num of images, counting the prediction as correctly labelled if the correct label is in the top 5 predictions, not just the top 1 prediction

Question 38

How does back propagation work?

Answer 38

- blaming individual weights for output, identified by calculating the loss - adapting weight and all the weights in following layers that impacted by bad weight - calculate gradient to update the networks weights

Question 39

Describe MSE (Mean Squared Error)

Answer 39

find sum of all the losses (using L2 loss), then average it over the number input images: MSE = 1/P * sum of (prediction - ground truth)2

Question 40

How do we use the MSE mean squared error minimise loss?

Answer 40

To minimise the MSE we use the gradient descent method. * Gradient descent finds the absolute minimum of a function. * It is especially useful for high-dimensional functions. * It iteratively minimises the neuron’s error by finding the gradient of the error surface in weight-space and adjusting the weights in the opposite direction.