Uploaded by Dr. Utku Köse

Introduction to Artificial Intelligence (Dr. Utku Kose)

advertisement
METU Statistics Club // METU, Ankara, Turkey
19.06.2021
METU Statistics Club // METU, Ankara, Turkey
 What is Artificial Intelligence?
 Sub-fields of Artificial Intelligence / Some Techniques
 Artificial Intelligence in Present Problems
 Near and Far Future of Artificial Intelligence
 Open Problems
 Suggestions
 Questions?
19.06.2021
19.06.2021
METU Statistics Club // METU, Ankara, Turkey
Content
 What is AI?
 Sub-fields / Some Techns.
 AI in Present Problems
 Near and Far Future of AI
 Open Problems
 Suggestions
 Questions?
1. WHAT IS AI?
 What is it?







Somehow a simulation of humans, animals, and the nature,
Technology based intelligent solutions,
A way for improving effectiveness and efficiency,
Science of the future,
A multidisciplinary science,
An interface between human and the machine,
Mathematical and logical algorithms.






An utopia?
A technological risk, threat?
Cause of extinction?
A science without human?
A science evolving by its own efforts?
A stoppable technology?
19.06.2021
METU Statistics Club // METU, Ankara, Turkey
Content
 What is AI?
1. WHAT IS AI?
 Sub-fields / Some Techns.
 AI in Present Problems
 Near and Far Future of AI
Static definition: A sub-field of Computer Science in which
research studies on simulating human thinking, behaviors
and natural dynamics are done.
 Open Problems
 Suggestions
 Questions?
Dynamic definition: Achieving ‘as much as possible’ of
performing human thinking, behaviors and natural
dynamics by computer-based systems.
«Science and engineering
of making intelligent
machines…»
John McCarthy
METU Statistics Club // METU, Ankara, Turkey
Content
 What is AI?
 Sub-fields / Some Techns.
 AI in Present Problems
 Near and Far Future of AI
 Open Problems
 Suggestions
 Questions?
2. SUB-FIELDS OF AI / SOME TECHNS.
19.06.2021
19.06.2021
METU Statistics Club // METU, Ankara, Turkey
Content
 What is AI?
2. SUB-FIELDS OF AI / SOME TECHNS.
 Sub-fields / Some Techns.
 AI in Present Problems
Data
Processing
Communication
 Near and Far Future of AI
 Open Problems
AI
 Suggestions
 Questions?
Cybernetics
Image
Electronics
Machine
Learning
Deep
Learning
Robotics
Data Mining
Signal
19.06.2021
METU Statistics Club // METU, Ankara, Turkey
Content
 What is AI?
 Sub-fields / Some Techns.
 AI in Present Problems
2. SUB-FIELDS OF AI / SOME TECHNS.
 For control, without learning: Fuzzy Logic
 Near and Far Future of AI
 Open Problems
 Suggestions
 Questions?
Crisp Logic
Lotfali Askar Zadeh
Fuzzy Logic
Crisp Logic
Fuzzy Logic
19.06.2021
METU Statistics Club // METU, Ankara, Turkey
Content
 What is AI?
 Sub-fields / Some Techns.
 AI in Present Problems
 Near and Far Future of AI
2. SUB-FIELDS OF AI / SOME TECHNS.
 Machine Learning
/ Deep Learning
Model of the
Problem
 Open Problems
 Suggestions
 Questions?
Training Issue
Which technique?
Which parameters?
19.06.2021
METU Statistics Club // METU, Ankara, Turkey
Content
 What is AI?
 Sub-fields / Some Techns.
2. SUB-FIELDS OF AI / SOME TECHNS.
 A very famous Machine L. technique: Artificial Neural Networks
 AI in Present Problems
.
.
 Near and Far Future of AI
 Open Problems
.
.
 Suggestions
 Questions?
ANC
.
.
BNC
.
.
.
.
.
.
ANC
19.06.2021
METU Statistics Club // METU, Ankara, Turkey
Content
 What is AI?
3. AI IN PRESENT PROBLEMS
 Sub-fields / Some Techns.
 AI in Present Problems
 Near and Far Future of AI
Train / Test
 Open Problems
 Suggestions
 Questions?
Data
Feature Ext.
Application
Direct /
Indirect Use
Solution
METU Statistics Club // METU, Ankara, Turkey
Content
 What is AI?
 Sub-fields / Some Techns.
 AI in Present Problems
 Near and Far Future of AI
 Open Problems
 Suggestions
 Questions?
2. SUB-FIELDS OF AI / SOME TECHNS.
 Convolutional Neural Networks
19.06.2021
19.06.2021
METU Statistics Club // METU, Ankara, Turkey
Content
 What is AI?
 Sub-fields / Some Techns.
 AI in Present Problems
2. SUB-FIELDS OF AI / SOME TECHNS.
 Optimization inspired by Evolution: Genetic Algorithms
 Near and Far Future of AI
Choosing good members,
eliminating others
 Open Problems
 Suggestions
 Questions?
Solution starting with
an initial population
Applying evolutional operations
according to solution flow
Continuing to solution
process with new
members
Repeating same steps till a stopping
criteria is met
19.06.2021
METU Statistics Club // METU, Ankara, Turkey
Content
 What is AI?
3. AI IN PRESENT PROBLEMS
 Sub-fields / Some Techns.
 AI in Present Problems
 Near and Far Future of AI
Optimization
 Open Problems
 Suggestions
 Questions?
Problem
Model
Objectives
Single
Multi
Continuous
Combinatorial
METU Statistics Club // METU, Ankara, Turkey
Content
 What is AI?
 Sub-fields / Some Techns.
 AI in Present Problems
 Near and Far Future of AI
 Open Problems
 Suggestions
 Questions?
3. AI IN PRESENT PROBLEMS
19.06.2021
19.06.2021
METU Statistics Club // METU, Ankara, Turkey
Content
 What is AI?
3. AI IN PRESENT PROBLEMS
 Sub-fields / Some Techns.
 AI in Present Problems
 Near and Far Future of AI
Pattern
Recognition
 Open Problems
 Suggestions
 Questions?
Prediction
Optimization
AI
Adaptive
Control
Discovery
19.06.2021
METU Statistics Club // METU, Ankara, Turkey
Content
 What is AI?
3. AI IN PRESENT PROBLEMS
 Sub-fields / Some Techns.
 AI in Present Problems
 Near and Far Future of AI
Data Types
Learning
Styles
Evaluation
Metrics
 Open Problems
 Suggestions
 Questions?
Input-Output
Supervised
(Class. /
Regress.)
MSE
Input (only)
Unsupervised
(Clustering)
Confusion
Matrix
Knowledge
(Rules)
Reinforcement
Statistical
Validations
SemiSupervised
METU Statistics Club // METU, Ankara, Turkey
Content
 What is AI?
 Sub-fields / Some Techns.
 AI in Present Problems
 Near and Far Future of AI
 Open Problems
 Suggestions
 Questions?
4. FUTURE OF AI
 AI in daily life,
 Robotic assistans,
 Unmanned machines,
 New jobs, expertises,
 Change of current jobs.
19.06.2021
METU Statistics Club // METU, Ankara, Turkey
Content
 What is AI?
 Sub-fields / Some Techns.
 AI in Present Problems
 Near and Far Future of AI
 Open Problems
 Suggestions
 Questions?
4. FUTURE OF AI
19.06.2021
METU Statistics Club // METU, Ankara, Turkey
Content
 What is AI?
 Sub-fields / Some Techns.
 AI in Present Problems
 Near and Far Future of AI
 Open Problems
 Suggestions
 Questions?
4. FUTURE OF AI
 Artificial General Intelligence
19.06.2021
METU Statistics Club // METU, Ankara, Turkey
Content
 What is AI?
 Sub-fields / Some Techns.
 AI in Present Problems
 Near and Far Future of AI
 Open Problems
 Suggestions
 Questions?
4. FUTURE OF AI
 Ambient Intelligence
19.06.2021
METU Statistics Club // METU, Ankara, Turkey
Content
 What is AI?
 Sub-fields / Some Techns.
 AI in Present Problems
 Near and Far Future of AI
 Open Problems
 Suggestions
 Questions?
4. FUTURE OF AI
 Explainable Artificial Intelligence
19.06.2021
METU Statistics Club // METU, Ankara, Turkey
Content
 What is AI?
 Sub-fields / Some Techns.
 AI in Present Problems
 Near and Far Future of AI
 Open Problems
 Suggestions
 Questions?
4. FUTURE OF AI
 Explainable Artificial Intelligence
19.06.2021
METU Statistics Club // METU, Ankara, Turkey
Content
 What is AI?
 Sub-fields / Some Techns.
 AI in Present Problems
 Near and Far Future of AI
 Open Problems
 Suggestions
 Questions?
4. FUTURE OF AI
 Artificial Intelligence Safety
19.06.2021
METU Statistics Club // METU, Ankara, Turkey
19.06.2021
Content
 What is AI?
 Sub-fields / Some Techns.
 AI in Present Problems
5. OPEN PROBLEMS
 New AI techniques, hybrid models…
 Near and Far Future of AI
 Choose of true parameter values, training data sets…
 Open Problems
 Solving new problems…
 Suggestions
 Robotics (traditional / soft-robotics)…
 Questions?
 How to train AI correctly?...
 Machine Learning…
 Supportive technologies…
 Openness in AI…
 Data (collecting, value…etc.)…
 Ethical and safety oriented issues…
 Existential risks…
 Bio-computer systems…
METU Statistics Club // METU, Ankara, Turkey
19.06.2021
Content
 What is AI?
 Sub-fields / Some Techns.
 AI in Present Problems
 Near and Far Future of AI
6. SUGGESTIONS
 Learn about philosophy of AI and its essentials…
 Learn algorithmic details of AI techniques…
 Open Problems
 Write codes first (do not use libraries directly)…
 Suggestions
 Learn Python, Java, OOP…
 Questions?
 Learn some about supportive technologies…
 Learn about linear alg., basic dif., non-linear trans., regression
essentials, numerical analysis, statistics…
 Learn some about mathematics and logics…
 Choose sub-field(s) to work in…
 Join to events, training programmes, online courses…
 Read about AI (news, blogs, videos, newsletters…etc.)…
 Read publications about AI…
 Follow famous scientists…
19.06.2021
METU Statistics Club // METU, Ankara, Turkey
Content
 What is AI?
7. QUESTIONS?
 Sub-fields / Some Techns.
 AI in Present Problems
 Near and Far Future of AI
 Open Problems
 Suggestions
 Questions?
«Thank you all for
listening!»

Assoc. Prof. Dr. Utku KÖSE
Suleyman Demirel University
Dept. of Computer Engineering
utkukose@gmail.com utkukose@sdu.edu.tr
http://www.utkukose.com
Download