Download Application of Artificial Intelligence in Today`s

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Enactivism wikipedia , lookup

Human–computer interaction wikipedia , lookup

Technological singularity wikipedia , lookup

Wizard of Oz experiment wikipedia , lookup

AI winter wikipedia , lookup

Ecological interface design wikipedia , lookup

Agent (The Matrix) wikipedia , lookup

Philosophy of artificial intelligence wikipedia , lookup

Existential risk from artificial general intelligence wikipedia , lookup

Intelligence explosion wikipedia , lookup

History of artificial intelligence wikipedia , lookup

Embodied cognitive science wikipedia , lookup

Ethics of artificial intelligence wikipedia , lookup

Transcript
Introduction
Acting Rationally
How to make a System Intelligent?
Application of Artificial Intelligence in Today’s
Software Industry
Subha Fernando, Dr.Eng. M.Eng, B.Sc(Special)(Hons)
[email protected]
Department of Computational Mathematics –Div. of Computational Intelligence
University of Moratuwa
18 December 2015
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Outline
1
Introduction
Intelligence
2
Acting Rationally
Rational Agents
BDI Agents
3
How to make a System Intelligent?
Interface Agent – Personalizing Web-based Tasks
State of AI Systems in Practice
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Intelligence
Introduction
What is AI?
It is the science and engineering of making intelligent machines,
especially intelligent computer programs. It is related to the similar
task of using computers to understand human intelligence.
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Intelligence
Introduction
What is AI?
It is the science and engineering of making intelligent machines,
especially intelligent computer programs. It is related to the similar
task of using computers to understand human intelligence.
Isn’t there a solid definition of intelligence that doesn’t depend on
relating it to human intelligence?
The problem is that we cannot yet characterize in general what
kinds of computational procedures we want to call intelligent. We
understand some of the mechanisms of intelligence and not others.
a
a
by John McCarthy, Stanford University
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Intelligence
What is Intelligence?
What is involved in Intelligence?
1
Ability to interact with the real world
to perceive, understand, and act
e.g., speech recognition and understanding and synthesis,
image understanding and ability to take actions, have an effect
(or the ability to change the status of the world)
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Intelligence
What is Intelligence?
What is involved in Intelligence?
1
Ability to interact with the real world
to perceive, understand, and act
e.g., speech recognition and understanding and synthesis,
image understanding and ability to take actions, have an effect
(or the ability to change the status of the world)
2
Reasoning and Planning
modeling the external world, given input
solving new problems, planning, and making decisions
ability to deal with unexpected problems, uncertainties
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Intelligence
What is Intelligence?
What is involved in Intelligence?
1
Ability to interact with the real world
to perceive, understand, and act
e.g., speech recognition and understanding and synthesis,
image understanding and ability to take actions, have an effect
(or the ability to change the status of the world)
2
Reasoning and Planning
modeling the external world, given input
solving new problems, planning, and making decisions
ability to deal with unexpected problems, uncertainties
3
Learning and Adaptation
we are continuously learning and adapting
our internal models are always being ”updated”,
e.g., a baby learning to categorize and recognize animals.
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Intelligence
Different Types of Artificial Intelligence
Acting Humanly- Modeling exactly how humans actually act
”The art of creating machines that perform functions that
require intelligence when performed by people”,
(Kurzweil,1990)
Ultimately to be tested by the Turing Test
This require physical interaction with environment:- Natural
language processing, Knowledge representation, Automated
reasoning, Machine learning.
Intelligence takes many forms, which are not necessarily best
tested this way.
Is it important that an intelligent system act like a human?,
because not all humans are intelligent.
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Intelligence
Different Types of Artificial Intelligence
Thinking Humanly- Modeling exactly how humans actually think
”The automation of activities that associate with human
thinking, activities such as decision-making, problem solving,
learning” (Bellman)
Goal is to build systems that function internally in some way
similar to human mind.
Traditional computer game players typically work much
differently than human players, s.t.Massive look-ahead,
minimal ”experience”.
People think differently in experience, ”big picture”, etc.
Requires detailed matching of computer behavior and timing
to detailed measurements of human subjects gathered in
psychological experiments.
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Intelligence
Different Types of Artificial Intelligence
Thinking Rationally- Modeling how ideal agents ”should think”
”The study of the computations that make it possible to
perceive, reason, and act” (Winston).
Approach firmly grounded in logic. i.e. formalize correct
reasoning using mathematical model (e.g. deductive
reasoning)
I.e., how can knowledge be represented logically, and how can
a system draw deductions?
Does not account for uncertainty about the world
General deductive inference is computationally intractable.
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Intelligence
Different Types of Artificial Intelligence
Acting Rationally- Modeling how ideal agents ”should act”.
”The branch of computer science that is concerned with the
automation of intelligent behavior” (Luger and Stubblefield).
The intelligent agent approach.
An agent is something that perceives, understand and acts
Emphasis is on behavior
Behavior is more scientifically testable than thought.
Rather than imitating humans trying to solve hard problems,
just try to solve hard problems
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Rational Agents
BDI Agents
A Software Agent
Definition (Agent)
An agent is a computer system that is situated in some
environment, and that is capable of autonomous action in this
environment in order to meet its design objectives.
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Rational Agents
BDI Agents
Acting Rationally
What is Acting Rationally?
An agent acts rationally if it selects the action that maximizes its
”utility” Or expected utility if there is uncertainty.
1
Decision theory:
Set of possible actions an agent can take.
Utility = gain to an agent for each action/state pair
2
Emphasis is on autonomous agents that behave rationally
(make the best predictions, take the best actions).
3
On average over time within computational limitations
(”bounded rationality”).
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Rational Agents
BDI Agents
Rational Agents
1
Basic Structure
Perceive the environment through sensors (Perceive)
Act upon the environment through actuators (Actions)
Act rational with respect to a performance measure (e.g. goal:
money, time, energy, utility)
2
Rational Behavior depends on:
Performance measures (goals), Precept sequences, Knowledge
of the environment, Possible actions
Algorithm 1 Pseudocode for Rational Agent
1:
2:
3:
4:
5:
6:
FUNCTION Agent(percept) RETURN action
static: memory {Agent’s Memory about the World}
memory :- update memory(percept,memory)
action :- choose action(memory)
memory :- update memory(memory, action)
RETURN action
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Rational Agents
BDI Agents
Belief-Desire-Intention Agents
Belief :- Agents’ belief about the surrounding environment.
Desires :- The states of affairs that are considered for achievement.
Intentions:- Intentions based on notion of persistent goal. A goal that
eventually becomes true, and believes that it is not currently true.
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Interface Agent – Personalizing Web-based Tasks
State of AI Systems in Practice
How to make a System Smarter or Intelligent?
Case Study: Interface Agent personalizing Web-based tasks
The volume of information available on the Web is constantly
growing. Due to this situation, users looking for documents
relevant to their interests need to identify them among all the
available ones. Intelligent agents have become a solution to assist
users in this task since they can retrieve, filter and organize
information on behalf of their users.
Environment :- Web (can access only text data and images)
Objectives :- Fulfill the Customer Needs
Possible Actions:- Read, Write, Speak, Search, Learn, Play Music,
Play Games, etc.
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Interface Agent – Personalizing Web-based Tasks
State of AI Systems in Practice
How to Fit into BDI Agent?
Case 1: Selling Product to a Customer who is bored
1
Percept : Emotional Status, text or voice queries.
Emotional Status - Emotion Detection using Image Processing
Voice Queries - Speech Recognition
Text Queries - Natural Language Processing
2
Percept will add beliefs to the Agents (After
understanding the Semantic of the Percepts)
+mood(bored). + need(book(book title = ”+artificial
intelligence+....”,. . . )).
3
Desires :-!satisfy(not happy(customer), ?customer(needs)),
!maximize profit(happy(customer), ?customer(needs))
4
Intention :- !satisfy(not happy(customer), ?customer(needs))
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Interface Agent – Personalizing Web-based Tasks
State of AI Systems in Practice
Making an Autonomous Agents
Autonomous
Capability of acting independently, exhibiting control over their
internal state.
Case 1: Selling Product to a bored Customer - How to make
the Agent Autonomous
1
How to generate a Plan – a recipe of Actions
making cust happy : [not happy(customer)]→
play back ground music, making cust happy (self=(success
=0.7(bored), =0.6(sad))).
making cust happy : [not happy(customer)] →
have chat(?context(book(book title= ”+artificial intelligence+”))),
making cust happy.self=(success =0.4(bored), =0.8(sad))).
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Interface Agent – Personalizing Web-based Tasks
State of AI Systems in Practice
Making an Autonomous Agents
Case 1: Selling Product to a bored Customer - How to make
the Agent Autonomous
1
2
Agent select a plan that will maximize the Expected Utility
(here it will be the success)
Agent can generate a plan randomly:
making cust happy : [not happy(customer)]→
play back ground music, have chat(?context(book(book title=
”+artificial intelligence+”)), making cust happy (self=(success
=new,
3
4
5
Autonomous is selecting set of actions according to your belief.
Environment constraints the actions that it can perform.
Execution of the plan (series of actions)should guarantee the
achievement of the intention
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Interface Agent – Personalizing Web-based Tasks
State of AI Systems in Practice
Learning and Adaptation
What is Learning and Adaption?
Machines should be capable of making changes to themselves over
time with the goal of improving their performance on the tasks
confronting them in a particular environment.
How to make them LEARN?
It can learn plans i.e. It fills
the self annotated observations
by trying out different plans.
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Interface Agent – Personalizing Web-based Tasks
State of AI Systems in Practice
Learning and Adaptation
How to make them LEARN? They can learn from the data
or Expand the Instance/Pattern Knowledge Bases.
SEED PATTERNS :
POS arg1 POS arg2 Cricketer
POS arg1 POS arg2 member of POS arg3 Cricket Team
INSTANCES :- Classified
Sangakkara is a Cricketer
Mahela is a member of Sri Lankan Cricket Team
INSTANCES :Non-Classified
Mahanama is a left hand batsman
Mahela is a right hand batsman
Sangakkara is a left hand batsman
PATTERNS :LEARNT
POS arg1 POS arg2 left hand batsman
POS arg1 POS arg2 right hand batsman
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Interface Agent – Personalizing Web-based Tasks
State of AI Systems in Practice
Learning and Adaptation
How to make them LEARN? They can learn from others’
experiences
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Interface Agent – Personalizing Web-based Tasks
State of AI Systems in Practice
Reasoning and Planning
What is Reasoning
Ability to model the external world, given input, and
Ability to deal with unexpected problems, uncertainties
1
Ability to model the external world
Understand the information conveyed through the Perception
(NLP, Speech Recognition, Image Analysis)
Predict the states of the outside environment using these
Observations (Machine Intelligence)
Make decision about the present state of the environment.
2
Ability to backtrack and reinitialize new plan when the
prediction fails.
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Interface Agent – Personalizing Web-based Tasks
State of AI Systems in Practice
Reasoning and Planning
making cust happy : [not happy(customer)] →
have chat(?context(book(book title= ”+artificial intelligence+”))),
making cust happy.self=(success =0.4(bored), =0.8(sad))).
Percept : mood(happy) → happy(customer)
Intention: maximize profit(happy(customer), ?customer(needs))
making profit : [happy(customer)] →
browse similar items(?context(book(book title= ”+artificial
intelligence+”)), booklist), orde items(booklist,sentiment order)
making profit : [happy(customer)] →
browse similar items(?context(book(book title= ”+artificial
intelligence+”)), booklist), orde items(booklist,sentiment order),
pop up(related books(booklist))
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Interface Agent – Personalizing Web-based Tasks
State of AI Systems in Practice
State of AI Systems in Practice
Image Reading by Convolutional Networks
The intuition of Convolutional
Networks
Try to Remember the following
Number:
1223334444
Redundancy and Organization
of local features are important
to reconstruct the number
again.
Very Important: Not all the
possible combinations of local
features reconstruct the number.
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Interface Agent – Personalizing Web-based Tasks
State of AI Systems in Practice
Image Classification and Localization by Convolutional Networks
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Interface Agent – Personalizing Web-based Tasks
State of AI Systems in Practice
Handwriting Synthesis by Long-Short-Term Memory (LSTM)
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Interface Agent – Personalizing Web-based Tasks
State of AI Systems in Practice
Handwriting Synthesis by Long-Short-Term Memory (LSTM)
http://www.cs.toronto.edu/~graves/handwriting.html
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Interface Agent – Personalizing Web-based Tasks
State of AI Systems in Practice
Simple Object Recognition by ConvNet + LSTM
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Interface Agent – Personalizing Web-based Tasks
State of AI Systems in Practice
Simple Object Recognition by ConvNet + LSTM
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Interface Agent – Personalizing Web-based Tasks
State of AI Systems in Practice
Autonomous Vehicles/Robots
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY
Introduction
Acting Rationally
How to make a System Intelligent?
Interface Agent – Personalizing Web-based Tasks
State of AI Systems in Practice
Academic Disciplines relevant to AI
Philosophy :- Logic, methods of reasoning, mind as physical
system, foundations of learning, language,rationality.
Mathematics:- Formal representation and proof, algorithms,
computation, (un)decidability, (in)tractability.
Probability/Statistics:- modeling uncertainty, learning from
data
Economics:- utility, decision theory, rational economic agents
Neuroscience:- neurons as information processing units.
Psychology/Cognitive Science:- how do people behave,
perceive, process cognitive information, represent knowledge.
Computer Engineering:- building fast computers
Control theory:- design systems that maximize an objectives
function over time
Linguistics:- knowledge representation, grammars
Subha Fernando, Dr.Eng
APP. OF AI. IN SOFTWARE INDUSTRY