Foundations

Question 1

Computer vision

Answer 1

Image detection and classification. Near realtime video analysis

Question 2

Computer vision ML model

Answer 2

Uses neural networks

Question 3

3 components of neural networks

Answer 3

Input layer: receives input data
Hidden layer: finds features in the data that have predictive power based on labels
Output layer: Generates the output

Question 4

CNN

Answer 4

Convolutional neural networks used in modern day computer vision. Provides for faster training

Question 5

Feature extraction

Answer 5

In CNNs this describes the process of how hidden layers extract different information from images

Question 6

Image classification

Answer 6

What’s in this image. Text detection, OCR and content moderation

Question 7

Object detection

Answer 7

More granular than image classification. How many objects or are there different classes of the object in the image

Question 8

Semantic segmentation

Answer 8

Goes down to the pixel level. Identifies which part of the image is the object in

Question 9

Activity recognition

Answer 9

Based around videos. Added time component. Detects changes that occur over time

Question 10

Machine learning (ML)

Answer 10

is a modern software development technique and a type of artificial intelligence (AI) that enables computers to solve problems by using examples of real-world data. It allows computers to automatically learn and improve from experience without being explicitly programmed to do so.

Question 11

supervised learning

Answer 11

every training sample from the dataset has a corresponding label or output value associated with it. As a result, the algorithm learns to predict labels or output values.

Question 12

unsupervised learning

Answer 12

there are no labels for the training data. A machine learning algorithm tries to learn the underlying patterns or distributions that govern the data.

Question 13

reinforcement learning

Answer 13

the algorithm figures out which actions to take in a situation to maximise a reward (in the form of a number) on the way to reaching a specific goal. This is a completely different approach than supervised and unsupervised learning.

Question 14

model

Answer 14

an extremely generic program, made specific by the data used to train it. A more technical definition would be that a machine learning model is a block of code or framework that can be modified to solve different but related problems based on the data provided.

Question 15

Model training algorithms

Answer 15

work through an interactive process where the current model iteration is analysed to determine what changes can be made to get closer to the goal. Those changes are made and the iteration continues until the model is evaluated to meet the goals.

Question 16

Model inference

Answer 16

is when the trained model is used to generate predictions.

Question 17

5 machine learning steps

Answer 17

1. Define the problem
2. Build the dataset
3. Train the model
4. Evaluate the model
5. Use the model

Question 18

labels

Answer 18

refers to data that already contains the solutions

Question 19

Supervised vs unsupervised learning

Answer 19

Labelled data is supervised. Unlabelled data is unsupervised

Question 20

clustering

Answer 20

Used in unsupervised learning. Used to determine if there are any naturally occurring groups in the data

Question 21

categorical label

Answer 21

Has a discrete value (Sunday, Monday etc). Often used in classification tasks

Question 22

continuous label (regression)

Answer 22

No discrete possibility, means you’re likely working with numerical data

Question 23

Discrete

Answer 23

A term taken from statistics referring to an outcome taking on only a finite number of values (such as days of the week).

Question 24

Using unlabelled data

Answer 24

you don’t need to provide the model with any kind of label or solution while the model is being trained.

Question 25

The Four Aspects of Working with Data

Answer 25

1. Data collection
2. Data inspection
3. Summary statistics
4. Data visualisation

Question 26

Data collection

Answer 26

SQL queries, web scraping etc. Perhaps run a model over the data to generate the labels. Question: Does the data you’ve collected match the machine learning task and problem you have defined?

Question 27

Data inspection

Answer 27

Quality of the data will affect how your model performs. . Look for outliers, missing or incomplete values and maybe use a pre-processor to transform the data into the correct format

Question 28

Summary statistics

Answer 28

check that your data is in line with the underlying assumptions of your chosen machine learning model.
Use mean, inner quartile range, standard deviation. This provides insight into the scope, scale, and shape of the dataset.

Question 29

Data visualization

Answer 29

Use this to see outliers and trends in the data.

Question 30

Impute

Answer 30

referring to different statistical tools which can be used to calculate missing values from your dataset

Question 31

Outliers

Answer 31

data points that are significantly different from others in the same sample

Question 32

Dataset splitting

Answer 32

How: Randomly split the data

1. Training dataset - most of the data is here. 80%
2. Testing dataset - used to evaluate the model against unseen data. i.e. how well the model will generalise to new data

Question 33

What does the model training algorithm do

Answer 33

The model training algorithm iteratively updates a model’s parameters to minimize some loss function.

Question 34

Model parameters

Answer 34

Model parameters are settings or configurations the training algorithm can update to change how the model behaves

Question 35

Loss function

Answer 35

A loss function is used to codify the model’s distance from this/its goal.

Question 36

The end-to-end training process is

Answer 36

1. Feed the training data into the model.
2. Compute the loss function on the results.
3. Update the model parameters in a direction that reduces loss.

continue to cycle through these steps until you reach a predefined stop condition. This might be based on a training time, the number of training cycles, or an even more intelligent or application-aware mechanism

Question 37

model selection

Answer 37

A process to determine which model to use

Question 38

Hyperparameters

Answer 38

settings on the model which are not changed during training but can affect how quickly or how reliably the model trains, such as the number of clusters the model should identify.

Question 39

machine learning frameworks

Answer 39

Practitioners often use machine learning frameworks that already have working implementations of models and model training algorithms. You could implement these from scratch, but you probably won’t need to do so unless you’re developing new models or algorithms.

Question 40

*Linear models

Answer 40

One of the most common models covered in introductory coursework, linear models simply describe the relationship between a set of input numbers and a set of output numbers through a linear function (think of y = mx + b or a line on a x vs y chart).

Question 41

*Tree-based models

Answer 41

Tree-based models are probably the second most common model type covered in introductory coursework. They learn to categorize or regress by building an extremely large structure of nested if/else blocks, splitting the world into different regions at each if/else block. Training determines exactly where these splits happen and what value is assigned at each leaf region.

Question 42

*Deep learning models

Answer 42

Extremely popular and powerful, deep learning is a modern approach based around a conceptual model of how the human brain functions. The model (also called a neural network) is composed of collections of neurons (very simple computational units) connected together by weights (mathematical representations of how much information to allow to flow from one neuron to the next). The process of training involves finding values for each weight.

Question 43

*neural network structures

Answer 43

FFNN: The most straightforward way of structuring a neural network, the Feed Forward Neural Network (FFNN) structures neurons in a series of layers, with each neuron in a layer containing weights to all neurons in the previous layer.
CNN: Convolutional Neural Networks (CNN) represent nested filters over grid-organized data. They are by far the most commonly used type of model when processing images.
RNN/LSTM: Recurrent Neural Networks (RNN) and the related Long Short-Term Memory (LSTM) model types are structured to effectively represent for loops in traditional computing, collecting state while iterating over some object. They can be used for processing sequences of data.
Transformer: A more modern replacement for RNN/LSTMs, the transformer architecture enables training over larger datasets involving sequences of data.

Question 44

*Machine Learning Using Python Libraries

Answer 44

For more classical models (linear, tree-based) as well as a set of common ML-related tools, take a look at scikit-learn. The web documentation for this library is also organized for those getting familiar with space and can be a great place to get familiar with some extremely useful tools and techniques.
For deep learning, mxnet, tensorflow, andpytorch are the three most common libraries. For the purposes of the majority of machine learning needs, each of these is feature-paired and equivalent.

Question 45

Model accuracy

Answer 45

Accuracy is the fraction of predictions a model gets right.

Question 46

*Log loss

Answer 46

seeks to calculate how uncertain your model is about the predictions it is generating. In this context, uncertainty refers to how likely a model thinks the predictions being generated are to be correct.

Question 47

Model inference

Answer 47

generate predictions on real-world problems using unseen data in the field

Question 48

*root mean square (RMS)

Answer 48

RMS can be thought of roughly as the “average error” across your test dataset, so you want this value to be low

Question 49

Hyperplane

Answer 49

A mathematical term for a surface that contains more than two planes.

Question 50

Plane

Answer 50

A mathematical term for a flat surface (like a piece of paper) on which two points can be joined by a straight line.

Question 51

stop words

Answer 51

A list of words removed by natural language processing tools when building your dataset. There is no single universal list of stop words used by all-natural language processing tools.

Question 52

data vectorization

Answer 52

A process that converts non-numeric data into a numerical format so that it can be used by a machine learning model

Question 53

k-means & k

Answer 53

a common cluster-finding model. k is how many clusters the model will try to find in your dataset.

Question 54

silhouette coefficient

Answer 54

A score from -1 to 1 describing the clusters found during modeling. A score near zero indicates overlapping clusters, and scores less than zero indicate data points assigned to incorrect clusters. A score approaching 1 indicates successful identification of discrete non-overlapping clusters.

Question 55

Bag of words

Answer 55

A technique used to extract features from the text. It counts how many times a word appears in a document (corpus), and then transforms that information into a dataset

Question 56

CNN

Answer 56

are a special type of neural network particularly good at processing images.

Question 57

Neural network

Answer 57

a collection of very simple models connected together.

1. These simple models are called neurons
2. the connections between these models are trainable model parameters called weights.