Ch.14 Artificial Intelligence

Question 1

AI

Answer 1

theory and development of IS that are capable of performing tasks that normally require human intelligence.

Question 2

Main goal of AI

Answer 2

build machines that mimic human intelligence.

Question 3

Difference between. strong AI & weak AI

Answer 3

Strong(general) is AI that matches or exceeds human int. Weak(narrow) performs useful specific functions that were once done by human intelligence to perform. e.g robots

Question 4

Technological advancements that led to advancements in artificial
intelligence:

Answer 4

Technological advancements that led to advancements in artificial
intelligence:
o Advancements in chip technology
o Big Data
o The Internet and cloud computing
o Improved algorithms
o Algorithm: a problem-solving method expressed as a finite sequence of steps.

Question 5

Machine Learning (ML)

Answer 5

An application of AI that provides systems with the ability to automatically learn and improve from experience without being explicitly programmed.

Question 6

Traditional vs ML

Answer 6

ML compares the output to the expected results while traditional is a combination of data that produces answers.

Question 7

Expert system Vs ML

Answer 7

ES requires human experts to provide knowlegde for the system and must be formally structured while ML don’t require human experts and learn from ingesting vast amounts of data.

Question 8

Where is the knowledge stored in ES

Answer 8

in the form of IF-THEN rules

Question 9

ML Bias

Answer 9

Underspecification: Essentially, even if a training process can produce a good model, it could still ultimately produce a poor model. The process will not know the difference, and neither will the developers until the model is employed in the real world.
-how developers approach a problem,
-the data used to train the system; data shift mismatch bet. data used to train and test the system. Hidden relationship in the data

Question 10

Types of ML

Answer 10

• Supervised; the system is given labelled input data and expected output results.
  -Semi supervised; combines small amt. of data with large amt. of unlabelled data during training
  -Unsupervised; searches for previously undetected patterns in a data set with no pre existing label &minimal human supervision.(clustering- mkt seg.)
  -Reinforcement; the system learns to achieve a goal in an uncertain, complex environment. (Trial and error)
  -Deep; is a subset of ML in which artificail neural networks learn from large amt. of data.

Question 11

Classification(type of problem where a system predicts a category for given data) &Regression

Answer 11

*Binary classification; classf. problems that have only 2 class labels
* Multi-class classification; more than 2 class labels
* Multi-label classification; 2 or more labels, where 1 or more lables can be predicted for each ex.
* Imbalanced classification; the no. of classes in each class is unequally distributed
Linear regression (continuous variables rather than classifying into
categories)
* Simple linear regression; x is used to predict the value of y
* Multiple linear regression; 2 or more x is used

Question 12

Difference bet. multi class and multi label

Answer 12

in multi-class classification,belongs to only 1 category- the classes are mutually exclusive (e.g., an email is either spam or not). However, in multi-label classification, each label represents a different classification task(more than 1 category).

Question 13

When is the best time to use unsupervised learning

Answer 13

when an organization does not have data on desired outcomes.

Question 14

What happens Reinforcement Learning

Answer 14

The system must determine how to perform the task to maximize the reward, beginning with totally random trials and finishing with sophisticated tactics

Question 15

Neural Network

Answer 15

is a set of virtual neurons, or nodes, that work in parallel to simulate the way the human brain works, although in a greatly simplified form.

Question 16

Node

Answer 16

has one or more weighted input connections, a bias, an activation function, and one or more output connections.

Question 17

Types of Nodes in Neural

Answer 17

• Input Layer Node
  -Hidden Layer; bc it is located btw input&output layers.
• Output layer

Question 18

What are the no. of hidden layers

Answer 18

Hyperparameter

Question 19

Neural Networks for Specific Applications

Answer 19

o Recurrent neural networks (RNNs)
Designed to access previous data such as sequential or time series data during input iterations.
Examples: moving a robotic arm, predicting time series
o Convolutional neural networks (CNNs)
Designed to separate areas of image inputs by extracting features to identify edges, curves and colour density and then recombine these inputs for classification and prediction.
Examples: facial recognition, natural language processing
o Generative adversarial networks (GANs)
Two neural networks compete with each other in a zero-sum game to segregate real data from synthetic
data.
Generator: learns to generate plausible data.
Discriminator: learns to distinguish the generator’s fake data from the real data.
Example: deep-space photography for inpainting

Question 20

Artificial Intelligence Applications

Answer 20

Computer vision
Natural language processing
Robotics
Speech recognition
Chatbots

Question 21

Artificial Intelligence in the Functional Areas

Answer 21

• AI in Accounting
  o Taxes
  o Auditing
• AI in Finance
  o Process automation
  o Security
  o Insurance and risk management
  o Algorithmic trading
• AI in Marketing
  o Improved lead scoring accuracy
  o Easier customer churn prediction
  o Improved audience insights
  o Intelligent marketing campaigns
• AI in Production/Operations Management
  o In the factory
  o In transportation
  o Security
• AI in Human Resources
  o Recruiting
  o Onboarding
  o Career pathing
• AI in Management Information Systems
  o Security
  o Server optimization
  o Service management
  o Software development