Wednesday, June 6, 2018

Introduction to Artificial Intelligence and Approaches






Introduction

Artificial intelligence (AI) is the intelligence of machine and the branch of computer science that aims to create it. AI textbooks define the field as “ the study and design of intelligent agents” where an intelligent agent is a system that perceives its environment and takes actions that maximise its chances of success. 




Artificial intelligence (Al) is the intelligence exhibited by machines or software. 

It is also the name of the academic filed of study with studies how to create computers and computer software that are capable of intelligent behaviour. 

John McCarthy, who coined the term in 1955, defines it as “ the science and engineering of making intelligent machines”.

Approaches

Cybernetics and Brain Simulation

Symbolic

Cognitive Simulation

Logic-based

“Anti-logic” or “Scruffy”

Knowledge-based

Sub-symbolic

Bottom-up, Embodied, Situated, Behaviour-based Or Nouvelle AI

Computational Intelligennce and Soft Computing

Statistical

Integrating the Approaches: Intelligent Agent Paradigm

Agent Architectures and Cognitive Architectures



Cybernetics and Brain Simulation


In the 1940s and 1950s, a number of researchers explored the connection between neurology, information theory, and cybernetics. 



Some of them built machines that used electronic network's to exhibit rudimentary intelligence, such as W. Grey Walter’s turtles and the Johns Hopkins Beast. 

Many of these researchers gathered for meetings of the Teleological Society at Princeton University and the ratio Club in England. 

By 1960, this approach was largely abandoned, although elements of it would be revived in the 1980s. 




Symbolic


When access to digital computers became possible in the middle 1950s, Al research began to explore the possibility that human intelligence could be reduced to symbol manipulation. 



The research was centred in three institutions: Carnegie Mellon University, Stanford and MIT, and each one developed its own style of research. 

John Haugeland named these approaches to AI “good old fashioned AI” or “ GOFAI”. 

During the 1960s, symbolic approaches had achieved great success at simulating high-level thinning in small demonstration programs. 

Approaches based on cybernetics or neural networks where abandoned or pushed into the background 

Researchers in the 1960s and the 1970s were convinced that symbolic approaches would eventually succeed in creating a machine with artificial general intelligence and considered this the goal of there field.



Cognitive Simulation

Economist Herbert Simon and Allen Newell studied human problem-solving skills and attempted to formalise them, and their work laid the foundations of the field of artificial intelligence, as well as cognitive science, operation research and management science. 



Their research team used the result of psychological experiments to develop programs that simulated the techniques that people used to solve problems. 


This tradition, centred at Carnegie Mellon University would eventually culminate in the development of the Soar architecture in the middle 1980s.


Logic-based


Unlike Newell and Simon, Jon McCarthy felt that machine did not need to simulate human thought, but should instead try to find the essence of abstract reasoning and problem solving, regardless of whether people used the same algorithms. 



His laboratory at Stanford( SAIL) focused on using formal logic to solve a wide variety of problems, including knowledge representation, planning and learning. 

Logic was also the focus of the work at the University of Edinburgh and elsewhere in Europe which led to development of the programming Augean Prolog and the Science of logic programming.


“Anti-logic” or “Scruffy”


Researchers at MIT (such as Marvin Minskly and Seymour Papert) found that solving difficult problems in vision and natural language processing required ad-hoc solutions- they argued that there was no simple and general principle (like logic) that would capture all the aspects of intelligent behaviour. Roger Schank described their “ anti-logic” approaches as “ Scruffy” ( as opposed to the “neat” Paradigms at CMU and Stanford). 

Commonsense knowledge based (such as Doug Lenat’s Cyc) are an example of “ Suruffy” Al, since they must be built by hand, one complicated concept at a time. 


Knowledge-based


When computers with large memories became available around 1970, researchers from all three traditions began to build knowledge into AI applications. 


This “ Knowledge revolution “ led to the development and deployment of expert systems ( introduction by Edward Feigenbaum), the first truly successful from of AI Software. 

The knowledge revolution was also driven by the realisation that enormous amounts of knowledge would be required by many simple AI applications. 



Sub-symbolic


By the 1980s progression symbolic AI seemed to stall and many believed that symbolic system would never be able to imitate all the processes of the human cognition, especially perception, robotics, learning and pattern recognition. 



A number of researchers began to look into “ sub-symbolic” approaches to specific AI problems.



Bottom-up, Embodied, Situated, Behaviour-based Or Nouvelle AI


Research from there related field of robotics, such as Rodney Brooks, Rejected symbolic AI and Focused on the basic engineering problems that would allow robots to move and survive. 



Their work revived the non-symbolic viewpoint of the early cybernetics researchers of the 1950s and reintroduced the use of control theory in AI. 

This Coincided with the development of the embodied mind thesis in the related field of cognitive science: the idea that aspects of the body (such as movement, perception and visualisation) are required for higher intelligence.



Computational Intelligence and Soft Computing


Interest in neural network’s and “ connection-ism” was revived by David Rumelhart and other in middle 1980s. 



Neural networks are an example of soft computing --- they are solutions to problems which cannot be solved with complete logical certainty, and where an approximate solution is often enough. 

Other soft Computing approaches to Al include fuzzy systems, evolutionary computation and many statistical tools. 

The application of soft computing to AI is studied collectively by the emerging discipline of computational intelligence.



Statistical


In the 1990s, AI researchers develop sophisticated mathematical tools to solve specific sub problems. 



These tools are truly scientific, in the sense that their results are both measurable and verifiable, and they have been responsible for many of AI’s recent successes. 

The shared mathematical language has also permitted a high level of collaboration with more established fields (like mathematics, economics or Operations research). 

Stuart Russell and Peter Novig describe this movement as nothing less then a “ revolution” and “ the victory of the neats.” 

Integrating the Approaches: Intelligent Agent Paradigm


An intelligent agent is a system that perceives its environment and takes action which maximise its changes of success. 



The simplest intelligent agents are programs that solve specific problems. 

More complicated agents include human beings and organisation of human beings (such as firms). 

The paradigm give researchers licences to study isolated problems and find solution that are both variable and useful, without agreeing on one single approach. 

An agent that solves a specific problem can use any approach that works – some agent are symbolic and logical, some are dsub-symbolic neural networks and others may use new approaches.



Agent Architectures and Cognitive Architectures


Researchers have designed systems to build intelligent systems out of interacting intelligent agents in a multi-agent system. 



A system with both symbolic and sub-symbolic components is a hybrid intelligent system, and the study of such systems is artificial intelligence system integration. 

A hierarchical control system provides a bridge between sub-symbolic AI at its lowest, reactive levels and traditional symbolic AI at its highest levels, where relaxed time constraints permit planning and world modelling. 

Rodney Brook’s subsumption architecture was an early proposal for such a hierarchical system.




Tools

Search and optimization

Logic

Probabilistic methods for uncertain reasoning

Classifiers and statistical learning methods

Neural networks

Deep neural networks

Control theory


Languages



Search and Optimisation


Many problems in AI can be solved in theory by intelligently searching thought many possible solutions; Reasoning can be reduced to performing a search. 



For example, Logical proof can be viewed as searching for a path that leads from premises to conclusions, where each step is the application of a interface rule. 

Planning algorithms search through trees of goals and sub-goals, attempting to find a path to target goal, a process call means-ends analysis. 

Robotics algorithms for moving limbs and grasping objects use local searches in configuration space. 

A very different kind of search came to prominence in the 1990s, based on the mathematical theory of optimisation 

These algorithms can be visualised as blind hill climbing: we begin the search at a random point on the landscape, and then, by jumps or steps, we keep moving our guess uphill, until we reach the top. Other optimisation algorithms are simulated annealing, beam search and random optimisation.


Logic


Logic is used for knowledge representation and problem solving, but it can be applied to other problems as well. 



Propositional or sequential logic is of statements which can be true or false. 

First-order logic also allows the use of quantifiers and predicates, and can express facts about objects, their properties, and their relations with other. 

Fuzzy logic, is a version of first-order logic which allows the truth of a statement to be represented as a value between 0 and 1, rather than simply True (1) or False (0). 

Default logic's, Non-Monotonic logic's and circumscription are forms of logic designed to help with default reasoning and the qualification problem. 

Several extensions of logic have been designed to handle specific domains of knowledge, such as: description logic's; situation calculus, event calculus and fluent calculus ( for representing events and time ); causal calculus; belief calculus; and modal logic's.



Probabilistic Methods for Uncertain Reasoning


Many problems in AI ( in reasoning, planning, learning, perception and robotic) require the agent to operate with incomplete or uncertain information. 



Bayesian networks are a very general tool that can be used for a large number of problems: reasoning( using the Bayesian inference algorithm), learning ( using the expectation-maximisation algorithm), planning (using decision networks) and perception(using dynamic Bayesian Networks). 

Probabilistic algorithms can also be used for filtering, prediction, smoothing and finding explanations for streams of data, helping perception systems to analyse processes that occur over time (e.g. hidden Markov models or Kalman Filter). 


Classifiers and Statistical Learning Methods


The simplest Al application can be divided into two types: classifiers (“ if shiny then diamond”) and controllers ( ‘ if Shiny them pick up”). 



Controllers do, however, also classify conditions before inferring actions, and therefore classification forms a centre part of many AI systems. 

Classifiers are functions that use pattern matching to determining a closest match. 

In supervised learning, each pattern belongs to certain predefined class. A class can be seen as decision that has to be made. 

A classifier can be trained in various ways; there are many statistical and machine learning approaches. 

The most widely used classifiers are the neural network, kernel methods such as the support vector machine, K-nearest neighbour algorithm, Gaussian mixture model naive Bayes classifier, and decision tree.



Neural Networks


A neural network is an interconnected group of nodes, akin to the vast network of neurons in the human brain. 



The study of artificial neural networks began in the decade before the fields of AI research was founded, in the work of Walter Pitts and Warren McCullough. 

Other important early researchers were Frank Rosenblatt. Who invented the perceptron and Paul Werbos who developed the back-propagation algorithm. 

The main categories of networks are a cyclic or feed-forward neural networks(where the signal passes in only one direction) and recurrent neural networks(which allow feedback) 

Neural networks can be applied to the problem of intelligent control ( for robotics) or learning, using such techniques as Hebbian learning and competitive learning. 



Deep Neural Networks


A deep neural network is an artificial neural network with multiple hidden layers of units between the input and output layers. 



Similar to shallow artificial neural networks, deep neural networks can model complex non-liner relationships. 

Over the last few years, advances in both machine learning algorithms and computer hardware have led to more efficient methods for training deep neural networks that contain many layers of non-linear hidden units and a very large output layer.



Control theory


Control theory, the grand child of cybernetics, has many important applications, especially in robotics. 



Languages


AI researchers have developed several specialised languages for AI research, including Lips and Prolog.



Problem Deducted Through AI

Deduction, Reasoning, Problem Solving.

Motion and Manipulation.

Natural Language Processing

Social Intelligence




Advantages


Can take on stressful and complex work that humans may struggle/can not do. 



Can complete task faster then a human can most likely. 

To discover unexplored things. i.e outer space 

Less errors and defects 

Function is infinite



Disadvantage


Lack the “human touch” 



Has the ability to replace human jobs 

Can malfunction and do the opposite of what they are programmed to do 

Can be misused leading to mass scale destruction



Applications


Gesture recognition 



Individual voice recognition 

Global voice recognition 

Facial expression recognition for interpretation of emotion and non verbal queues. 

Robot navigation.


References:

Wikipedia Website 
Artificial Intelligence : a modern approach | Stuart J. Russell and Peter Norving | Third Edition




No comments:

Post a Comment

Google+ Followers