Artificial Intelligence 101

Question 1

Definition of AI (Artificial Intelligence)

Answer 1

The science and engineering of making computers behave in ways that, until recently, we thought required human intelligence.

Question 2

Neural Network

Answer 2

A computer system or series of algorithms designed to function like the human brain. It’s based on a collected of connected units, called nodes or neurons.

Question 3

Artificial Neurons

Answer 3

A mathematical function that seeks to mimic the behavior of a biological neuron in the brain. It is composed of a set of weighted inputs. Altogether, artificial neurons make up a neural network.

Question 4

Machine Learning

Answer 4

A subset of AI where the focus is on computer algorithms that can learn from data. These algorithms can

Question 5

Deep Learning

Answer 5

A subset of Machine Learning where the focus is on multi-layered neural networks that can learn from large datasets.

Question 6

Name the three main types of machine learning techniques

Answer 6

Supervised Learning, Unsupervised Learning, & Reinforcement Learning

Question 7

Classification and Regression are methods of:

Answer 7

Supervised Learning

Question 8

Clustering and Association are methods of:

Answer 8

Unsupervised Learning

Question 9

Giving feedback to an algorithm when it does something right or wrong based on a discrete outcome. It can be real-time or offline.

Answer 9

Reinforcement Learning

Question 10

Classification

Answer 10

the problem of identifying to which of a set of categories a new observation belongs, on the basis of a training set of data containing observations (or instances) whose category membership is known

Question 11

Regression

Answer 11

a technique to predict continuous values, often quantities such as amounts or sizes with a task of approximating a mapping function

Question 12

Clustering

Answer 12

task of dividing data points or population into a number of groups

Question 13

Association

Answer 13

method for discovering interesting relations between variables in large databases

Question 14

Why is AI so relevant NOW?

Answer 14

Because our current level of compute power, data availability, and low cost are lining up to allow us to explore AI now better than ever.

Question 15

This type of learning maps an input to an output

Answer 15

Supervised learning

Question 16

_______ is tied to numeric outcomes

Answer 16

Regression

Question 17

_______ is tied to categorical or “class” outcomes

Answer 17

Classification

Question 18

Provide an example of supervised learning use cases:

Answer 18

Any of the following: Image classification, natural language processing, sentiment analysis, face recognition, optical character recognition, machine translation, audio transcription, event detection

Question 19

This type of learning finds patterns in data

Answer 19

Unsupervised learning

Question 20

Discrete data

Answer 20

Data that you can count and it has a finite amount

Question 21

Continuous data

Answer 21

(often numerical) data that takes a large range of values

Question 22

What would be the best type of algorithm to estimate the price of clothing?
(Classification, Regression, or Clustering)

Answer 22

Regression–Since the price data is continuous (in dollars and cents), and you have a known value that you want to estimate this is a supervised, regression task

Question 23

What does HITL mean?

Answer 23

“Human in the Loop” human-moderator or data annotator that can help with quality control of a product.

Question 24

What determines the decision-making model?

Answer 24

Weights and threshold

Question 25

What can AI do really well right now?

Answer 25

Specific, narrow tasks & learning from large volumes of unstructured data

Question 26

What can AI do fairly well right now, similar to most humans?

Answer 26

Optical character recognition, classification of images, handwriting recognition, facial recognition

Question 27

What is difficult for AI to do well right now?

Answer 27

Captions & visual descriptions of imagery, various robotics tasks (stable bipedal locomotion), general speech recognition, complex logical reasoning, tasks without context like translation & explanations

Question 28

What are the 5 main steps in project management for AI?

Answer 28

1. Business problem
2. Data
3. Model building
4. Deploy & measure
5. Active Learning & Tuning

Question 29

What is SCRUM?

Answer 29

A framework for prototyping and improving on product ideas and helps teams work together

Question 30

True/False: It’s typically recommended that you start with the problem, not the data.

Answer 30

True! It’s typically recommended that you start with the problem, not the data.

Question 31

What should you include in your project statement?

Answer 31

1. What problem we are solving
2. How AI will add value
3. What data are needed
4. The scope
5. How we measure success

Question 32

What improves your chances of success when defining your business problem?

Answer 32

Your business problem should be clear and specific, narrowed down as much as possible. Your project should deliver specific, measurable outcomes. Start with a business case so that you can evaluate your options for success.