LEARNING

Question 1

provides a computer with data, rather than explicit instructions. Using these data, the computer learns to recognize patterns and becomes able to execute tasks on its own.

Answer 1

machine learning

Question 2

a task where a computer learns a function that maps inputs to outputs based on a dataset of input-output pairs.

Answer 2

supervised learning

Question 3

This is a task where the function maps an input to a discrete output.

Answer 3

classification

Question 4

algorithm that, given an input, chooses the
class of the nearest data point to that input

Answer 4

nearest-neighbor classification

Question 5

algorithm that, given an input, chooses the
most common class out of the k nearest
data points to that input

Answer 5

k-nearest-neighbor classification

Question 6

drawback of k-nearest

Answer 6

using a naive approach, the algorithm will have to measure the distance of every single point to the point in question, which is computationally expensive

Question 7

how to optimize k-nearest

Answer 7

by using data structures that enable finding neighbors more quickly or by pruning irrelevant observations.

Question 8

drawback of perceptron learning

Answer 8

data are messy, and it is rare that one can draw a line and neatly divide the classes into two observations without any mistakes

Question 9

perceptron learning

Answer 9

goal of the algorithm is to find the best weight vector, when the algorithm encounters new data it updates the current weights.

Question 10

perceptron learning rule

Answer 10

for each data point, we adjust the weights to make our function more accurate.

Question 11

sequences of numbers

Answer 11

vector

Question 12

The weights and values in Perceptron Learning are represented using?

Answer 12

vectors

Question 13

unable to express uncertainty, since it can only be equal to 0 or to 1.

Answer 13

hard treshold

Question 14

uses a logistic function which is able to yield a real number between 0 and 1, expressing confidence in the estimate

Answer 14

soft treshold

Question 15

they are designed to find the maximum margin separator

Answer 15

Support Vector Machines

Question 17

A boundary that maximizes the distance between any of the data points

Answer 17

maximum margin separator

Question 18

Benefit of Support Vector Machines

Answer 18

they can represent decision boundaries with more than two dimensions, as well as non-linear decision boundaries

Question 19

this function gains value when the prediction isn’t correct and doesn’t gain value when it is correct

Answer 19

loss function

Question 20

supervised learning task of learning a function mapping an input point to a continuous value

Answer 20

regression

Question 21

functions that can be used when predicting a continuous value

Answer 21

L1 and L2 loss functions

Question 22

L2 Loss Function Formula

Answer 22

(actual - predicted)^2

Question 22

L1 Loss Function Formula

Answer 22

|actual - predicted|

Question 23

L1 vs L2

Answer 23

L₂ penalizes outliers more harshly than L₁ because it squares the the difference

Question 24

when a model fits the training data so well that it fails to generalize to other data sets.

Answer 24

overfitting

Question 25

process of penalizing hypotheses that are more complex to favor simpler, more general hypotheses.

Answer 25

regularization

Question 26

used to avoid overfitting

Answer 26

regularization

Question 26

a constant that we can use to modulate how strongly to penalize for complexity in our cost function

Answer 26

Lambda (λ)

Question 27

A technique for testing whether a model is overfitted by splitting data in two (testing and training set)

Answer 27

holdout-cross validation

Question 28

downside of holdout cross validation

Answer 28

we don’t get to train the model on half the data, since it is used for evaluation purposes.

Question 29

we divide the data into k sets. We run the training k times, each time leaving out one dataset and using it as a test set. We end up with k different evaluations of our model, which we can average and get an estimate of how our model generalizes without losing any data.

Answer 29

k-Fold Cross Validation

Question 30

Reinforcement Learning as a Markov Decision Process

Answer 30

- Set of states S - Set of actions Actions(S) - Transition model P(s’ | s, a) - Reward function R(s, a, s’)

Question 30

given a set of rewards or punishments, learn what actions to take in the future

Answer 30

reinforcement learning

Question 31

method for learning a function Q(s,a), outputs an estimate of the value of performing action a in state b

Answer 31

Q Learning

Question 32

Q-Learning Process

Answer 32

- The model starts with all estimated values equal to 0 (Q(s,a) = 0 for all s, a). - When an action is taken and a reward is received, the function does two things: 1. It estimates the value of Q(s, a) based on current reward and expected future rewards, and 2. Updates Q(s, a) to take into account both the old estimate and the new estimate.

Question 33

an algorithm that completely discounts the future estimated rewards, instead always choosing the action a in current state s that has the highest Q(s, a).

Answer 33

Greedy Decision-Making

Question 34

Explore vs. Exploit tradeoff

Answer 34

A greedy algorithm always exploits, taking the actions that are already established to bring to good outcomes. However, it will always follow the same path to the solution, never finding a better path. Exploration, on the other hand, means that the algorithm may use a previously unexplored route on its way to the target, allowing it to discover more efficient solutions along the way.

Question 35

In this type of algorithm, we set ε equal to how often we want to move randomly. With probability 1-ε, the algorithm chooses the best move (exploitation). With probability ε, the algorithm chooses a random move (exploration).

Answer 35

ε (epsilon) greedy algorithm

Question 36

allows us to approximate Q(s, a) using various other features, rather than storing one value for each state-action pair

Answer 36

function approximation

Question 37

given input data without any additional feedback/information, learn patterns.

Answer 37

unsupervised learning

Question 38

organizing a set of objects into groups in such a way that similar objects tend to be in the same group

Answer 38

clustering

Question 39

applications of clustering

Answer 39

- Genetic Research - Image Segmentation - Market Research - Medical Imaging - Social Network Analysis

Question 41

algorithm for clustering data based on repeatedly assigning points to clusters and updating those clusters’ center

Answer 41

k-means Clustering

Question 42

Categorize the following: A social network’s AI uses existing tagged photos of people to identify when those people appear in new photos.

Answer 42

This is an example of supervised learning

Question 43

Imagine a regression AI that makes the following predictions for the following 5 data points. What is the total L2 loss across all of these data points (i.e., the sum of all the individual L2 losses for each data point)? The true output is 2 and the AI predicted 4. The true output is 4 and the AI predicted 5. The true output is 4 and the AI predicted 3. The true output is 5 and the AI predicted 2. The true output is 6 and the AI predicted 5.

Answer 43

16

Question 44

If Hypothesis 1 has a lower L1 loss and a lower L2 loss than Hypothesis 2 on a set of training data, why might Hypothesis 2 still be a preferable hypothesis?

Answer 44

Hypothesis 1 might be the result of overfitting.

Question 45

In the ε-greedy approach to action selection in reinforcement learning, which of the following values of ε makes the approach identical to a purely greedy approach?

Answer 45

ε = 0