UCCD2063

Artificial Intelligence Techniques

Unit 1:
Introduction to Artificial Intelligence (AI)
 Outline
• Introduction
• What is AI
• History of AI
• Applications of AI
• Limitations of AI
• Course Outlines
 What is AI?

Sci-Fi AI

3
 What is AI?

“ Artificial intelligence (AI) is the science and

engineering of making intelligent machines,
especially intelligent computer programs.”
[John McCarthy, father of AI]

What is AI?
(https://www.youtube.com/watch?v=uMzUB89uSxU)

4
Three Types of AI

Machine Machine Machine

Learning Intelligence Consciousness

5
 History of AI
▪ 1950—70: Early AI, great expectations
• 1950: Alan Turing developed the Turing
▪ 1980s-late—90s: Second AI Winter
test to test machine intelligence
• 1950s: Early AI programs encoded in logic, ▪ 1990-2010: Machine Learning
including Samuel's checkers program, • Many learning models emerge:
Gelernter's Geometry Engine neural network, SVM, decision tree,
• 1956: Dartmouth meeting: “Artificial knn,...
Intelligence” adopted • Use of probability to model
• Problem: real world is too complex, search uncertainty
space grew exponentially for logical • AI Spring!
reasoning, outpacing hardware
▪ 2012-present: Rise of Deep Learning
▪ 1960s-late—70s: First AI Winter • 2012: AlexNet beats previous
benchmark on the ImageNet
▪ 1970—90: Knowledge-based AI competition
• 1970—80s: Expert systems: elicit specific • Neural networks gets deeper and
domain knowledge from experts in the larger (trillion parameters)
form of if-then rules. • Availability of very large datasets and
• Focused on building narrow practical fast GPU processor
systems in targeted domains. First real • Data will drive future discoveries and
application impacted the industry. alleviate the complexity in AI
• Problem: rules couldn't handle uncertainty
of the real world, too complex to maintain
6
 Evolution of AI Technique

Logic, search, if-then rule

Decision tree, SVM,
neural networks Deep neural networks

7
 Evolution of AI Technique

Rule-based
Expert System

Manual

Machine
Learning

Learn from features

Deep
Learning

Learn directly from data

8
 Machine Learning vs Deep Learning

Feature

Vector

Note: This course covers machine learning and traditional AI

techniques such as search and statistical inference. Deep
learning is covered in UCCD3074.
9
10
 Application of AI – Finance

Finance Technologies
▪ Input data: Financial data (e.g. statements, transactions)
▪ Output:
• Credit approval
• Fraud detection
• Sale prediction
• Inventory forecasting, …

11
 Applications of AI – Computer Vision
Visual Object Detection
▪ Input data: images
▪ Output: presence/absence of
particular objects in the
image

Video Surveillance
▪ Input data: video
▪ Output:
• Human activity detection
• Abnormal event detection

Medical Imaging
▪ Input data: X-ray images
▪ Output: presence/absence of
Tumor in the image
12
 Application of AI – Voice Recognition

Speech Technologies (e.g., Siri, Alexa)

▪ Input data: audio waveform
▪ Output:
• Text characters
• Automatic speech recognition (ASR)

13
Application of AI – Natural Language Processing
Language Processing Technologies
▪ Input data: text
▪ Output:
▪ Machine translation
▪ Question answering, personal assistant
▪ Text classification, spam filtering, etc.

14
 Application of AI – Autonomous Cars
▪ Autonomous Cars
• Input data: sensor data, video
camera, LIDAR system
• Output: Driving Control (steering
rotation, brake level, accelerator
level)

15
 Application of AI – more!

Healthcare
Manufacturing

Games
Route Planning
16
 Generative AI
▪ Typical machine-learning models learn to make a prediction based on
input data.
▪ Generative AI can be thought of as machine-learning models can
generate new content, such as text, images, music, or even videos,
similar to the data they were trained on, rather than making a
prediction about a specific dataset.
▪ These models are trained on enormous amounts of data to learn the
patterns, structures, and styles contained in the training dataset.

17
 Overview of AI Techniques

▪ AI covers a range of techniques that enable a computer

to do somethings that appears to be intelligent, such as:
• Search – search for solutions to some kind of problem
• Knowledge reasoning – drawing conclusion with logical
reasoning
• Optimization – finding the optimal solution to a problem
• Uncertainty – probabilistic (statistical) inference
• Machine learning and deep learning – learning from data

18
 Course Outlines

Markov Search
Machine Probabilistic Decision Problem
Learning Inference Process
Constraint
Deep Bayesian Adversarial Satisfaction
Learning Networks Search Problem Logic

Data Computation
(Intelligence from data) (Intelligence from computation)

Red = topics covered in this course 19

 Next:

The Fundamentals of Machine Learning