Artificial Intelligence Mr Arthur
Aims of Lesson 1 What is Intelligence?? Why is there so much confusion to its meaning
What is Intelligence???? “Intelligence is the name we give to the data processing activity of entities which respond to information with behaviour which appears to be intended to be optional with respect to pre-set goals” “Intelligence involves Knowing and  Choosing”
Intelligence?? Many researchers feel that intelligence has the following features The Ability to: Learn and adapt from experience. Use knowledge to make decisions. Problem solve. Show Creativity Handle and manipulate language “ Artificial intelligence is concerned with building machines that can act and react appropriately, adapting their response to the demands of situation.”
Confusion to its Meaning?? It is a subject which is developing all the time,  It is a subject which involves computer scientists, biologists, psychologists etc and they all have a different perspective  It is difficult to define human intelligence, let alone artificial intelligence.
Aims of Lesson 2 Test for Identifying Intelligence Early Developments in AI Game Playing
Testing Intelligence Turing Test Alan Turing, a British mathematician developed a test in 1950 to determine if a program was intelligent:  It has the following features: A human tester was connected to 2 terminals and asked a series of questions.  One of the terminals had another human making the responses and the other used a computer program to make the responses If the human tester could not distinguish between the human responder and the program response then the software was said to be intelligent.
Problems with Turing Test It needed a fairly limited problem domain i.e. the area of the knowledge had to be small. Did the program pretend to forget things or make errors in order to fool the human tester?
Early Developments in AI (1940-65) In the beginning the focus of AI research was on modelling the human brain.  Research shifted to using games like noughts and crosses, chess etc to create “AI” systems These games were effective as they had limited problem domains. The games also had a number of rules that were easy to define.
Aims of Lesson 3 Last Lesson Definition of intelligence Why can experts not agree on the definition of AI The Turing Test Early Developments in AI = Game Playing Today’s Lesson Language Processing ELIZA SHRDLU PARRY Chatterbots
Language Processing (1965-1975) In 1965 Researchers agreed that game playing programs could not pass the Turing test The focus shifted to language processing ELIZA (1966) 1 st  language processing program Responded to users inputs by asking questions based on previous responses
Language Processing (1965-1975) PARRY (1972) Parry modelled a conversation with a paranoid person This seems odd but the program was created by a psychiatrist  SHRDLU (1970) The program could interpret verbal commands to move coloured blocks "Move the red block behind the green one"  It could understand and carry out the instruction Chatterbots??
Aims of Lesson 4 Last Lesson Definition of intelligence Why can experts not agree on the definition of AI The Turing Test Early Developments in AI = Game Playing Language Processing ELIZA SHRDLU PARRY Chatterbots Today’s Lesson Knowledge representation Semantic Nets  Logic programming List Prolog Developments in Hardware AI questions on Forum
Knowledge Representation Semantic nets Diagrams which show how the information in a system is interlinked They can get very large and out of hand if you try to model a lot of knowledge
Knowledge Representation Logic Programming Implemented using declarative languages. Declarative Languages They have no fixed start or end point Consist of collection of facts and rules Use goal directed problem solving Prolog/Lisp
Knowledge Representation Early AI developments were hampered by the programming techniques used human conversations cannot easily be represented by an algorithm  A framework for representing knowledge' was a crucial moment for AI  semantic nets logic programming
Aims of Lesson 5 Last Lesson Definition of intelligence Why can experts not agree on the definition of AI The Turing Test Early Developments in AI = Game Playing Language Processing ELIZA SHRDLU PARRY Chatterbots Knowledge representation Semantic Nets  Logic programming List Prolog Today’s Lesson Developments in Hardware Applications of AI Intro to Natural Language Processing Uses of NLP NLP Steps
Developments in Hardware Faster processors More memory Increased backing storage capacity Computers being developed with multiple processors (parallel processing where a task can be processed simultaneously on different processors)
Uses of Natural Language Processing Customer query lines Vue cinema Bank of Scotland O2 Language translation e.g. English to French  Speech-driven word processors Command and control systems – giving verbal instructions to save, print etc
Problems with NLP Ambiguity of words and phrases I saw a man eating fish I saw a man hitting a boy with a stick Changing English language i.e. bouncebackability Accents and dialects i.e. I ken what you are talking about!! Similar sounding words and phrases e.g. furry boots are you fae?? Inconsistencies with grammar
Aims of Lesson 6 Last Lesson Definition of intelligence Why can experts not agree on the definition of AI The Turing Test Early Developments in AI = Game Playing Language Processing ELIZA SHRDLU PARRY Chatterbots Knowledge representation Semantic Nets  Logic programming List Prolog Developments in Hardware Applications of AI Intro to Natural Language Processing Today’s Lesson Steps in the NLP process Moral, Legal issues with NLP
Main Stages in NLP 1. Speech Recognition 2. Natural  Language  Recognition 3. Natural Language Generation 4. Speech Synthesis
NLP Process Speech Recognition Get sound input using microphone, sample and convert to digital data Segment the sound into recognisable sounds (phonemes)  Natural Language Recognition Put these sounds together to form words, sentences etc “ Understand" the meaning of these sentences  Natural Language Generation   Create an appropriate response  Speech Synthesis Convert the response into intelligible sound output
Moral and Legal Issues Moral NLP has benefited blind or disabled users who could not access computers  In military applications, it could be used in weapon systems Legal Mistranslations leading to commercial problems Possible wrong diagnosis
Aims of Lesson 7 Last Lesson Definition of intelligence Why can experts not agree on the definition of AI The Turing Test Early Developments in AI = Game Playing Language Processing ELIZA SHRDLU PARRY Chatterbots Knowledge representation Semantic Nets  Logic programming List Prolog Developments in Hardware Applications of AI Natural Language Processing Today’s Lesson Vision Systems Applications of Vision Systems Vision System Process
Vision Systems The aim is to develop systems which can see, make sense of what they see, and react accordingly.  Uses Security systems, recognising faces at airports Inspection of manufactured goods judging quality of production Vision systems on automated cars Visual stock control systems Control of AGVs Artificial Visual Sensing for the blind
Vision Systems Image Acquisition Signal Processing Edge detection
Vision Systems Object recognition (Pattern Matching) Its a Tin of Soup! Image Understanding
Vision System Process Image acquisition   Capturing the image and converting to digital Signal processing improving the resolution, removing distortion Edge detection   Identifying edges of image Object recognition Comparing the objects with known objects.   Image understanding How do objects relate to each other? What do they mean?
Aims of Lesson 8 Last Lesson Applications of AI Natural Language Processing Uses Problems Steps in NLP process Moral, Legal issues Vision Systems Uses 5 steps in process Today’s Lesson Expert Systems Examples in everyday life Advantages of Expert Systems Disadvantages of expert systems
Expert Systems An expert system is a computer program which uses a knowledge base of facts and rules populated by a human expert The expert system will give reliable advice on a limited area of expertise, and can interact and explain it reasoning to the user (justification facilities) Examples MYCIN to diagnose blood disorders Legal advice Chemical analysis, DENDRAL was designed to identify unknown substances  Car mechanic expert system BABY used to monitor premature babies
Advantages of Expert Systems Available to access 24/7 Expert System will never get tired It will never have an “Off day” You can have multiple copies of the expert system Can combine the knowledge of many experts No human emotion involved No possibility of expert system retiring No barrier due to poor communication skills
Disadvantages of Expert Systems No common sense is used If the expert system makes an error who is repsonsible? Developer? Company using it? Less trust in expert system?
Aims of Lesson 9 Last Lesson Applications of AI Natural Language Processing Uses Problems Steps in NLP process Moral, Legal issues Vision Systems Uses 5 steps in process Expert Systems Examples in everyday life Advantages of Expert Systems Disadvantages of expert systems Today’s Lesson Expert Systems Parts of an Expert System Expert System Shells Moral/Legal issues
Parts of an Expert System User Interface   This allows the user to input information into the Expert System and get responses in a language that the user understands  Knowledge base   Consists of all the facts and rules about the topic This information will have been provided by a  domain experts Converted into an appropriate form by a  knowledge engineer. Inference engine ,  This searches the knowledge base, comes to conclusions and  generates questions to the user
Expert System Shells An expert system shell is an expert system with no knowledge base.  It consists only of a user interface and an inference engine.  EMYCIN -  The interface was so good on this expert system that the company sold this as a shell and the user added the knowledge base
Legal and Moral Issues Legal If the system is wrong, who do you blame? Human expert, company who bought it, company who created it, the programmer Moral Would you like a machine to diagnose serious health problems Expert system wouldn't make a decision on what is morally correct, it would only use facts and rules It wouldn’t consider feelings Is it right to potentially make human experts redundant
Aims of Lesson 9 Last Lesson Applications of AI Natural Language Processing Uses Problems Steps in NLP process Moral, Legal issues Vision Systems Uses 5 steps in process Expert Systems Examples in everyday life Advantages of Expert Systems Disadvantages of expert systems Parts of an Expert System Expert System Shells Moral/Legal issues Today’s Lesson Intelligent Robots Smart Embedded Technology
Intelligent Robots Robots can be considered intelligent when they go beyond simple sensors and feedback  (dumb robots),  and display some further aspect of human-like behaviour Vision Systems The ability to learn and improve performance Robot that can walk rather than on wheels NLP response Examples Bomb disposal  Deep sea exploration and rescue  Space exploration Honda’s Asimo
Practical Problems with Intelligent Robots Provide a power supply? Implement effective 3D vision system? Implement mobility?  Providing adequate processing power? Manipulating objects?
Smart Embedded Technology SET is where a processor is inside another device, making the device “smart” e.g. it can make decisions and learn from experiences Examples  The car that can park itself without you The fridge that knows when food orders are running low and orders in more food Car engine control systems which monitor performance and inform the user when maintenance is required The tv system that works out what you might enjoy watching from past viewing patterns and records it when you are out
Aims of Lesson 10 Last Lesson Applications of AI Natural Language Processing Uses Problems Steps in NLP process Moral, Legal issues Vision Systems Uses 5 steps in process Expert Systems Examples in everyday life Advantages of Expert Systems Disadvantages of expert systems Parts of an Expert System Expert System Shells Moral/Legal issues Intelligent Robots Smart Embedded Technology Today’s Lesson Intro to Artificial Neural Systems
Artificial Neural Systems ANS is an approach to AI where the developer attempts to model the human brain The human brain is made up of billions of connections called Neutrons.  A Neutron will “fire” when it gets enough input from other Neutrons.  This firing of Neutrons is how the brain sends signals to the muscles
ANS Continued AI researchers have modelled this process using a network of artificial neutrons (perceptron).  If you connect a lot of these perceptrons together you can get some fairly complex decisions being made, which might rely on dozens of inputs.
ANS Continued Simply speaking, a neural network is an electronic model of the brain consisting of many interconnected simple processors. Neural networks can either be built in hardware as hard wired circuitry or be implemented in software.
Aims of Lesson 10 Last Lesson Applications of AI Natural Language Processing Uses Problems Steps in NLP process Moral, Legal issues Vision Systems Uses 5 steps in process Expert Systems Examples in everyday life Advantages of Expert Systems Disadvantages of expert systems Parts of an Expert System Expert System Shells Moral/Legal issues Intelligent Robots Smart Embedded Technology Intro to Artificial Neural Systems Today’s Lesson Layers 3 basic stages of ANS creation Applications of ANS Advantages/Disadvantages of ANS
ANS Layers ANS are made up of layers with input, internal (hidden) and output layers Perceptrons in the input layer react to different stimuli which triggers perceptrons in the hidden layers If the combined output from the hidden layers passes a certain threshold, it will pass an output signal You can therefore make fairly complex decisions based on dozens of separate inputs
Stages on Creating ANS Initial Setup   The network is created and the initial values (weights) are set for the input layer, the hidden layers and the output layers.  Training the Network A number of trials are run, to test the accuracy of the original threshold values.  The internal weights change, or rebalanced, so that the network gives the correct outputs.  These changes may be done manually or the network may rebalance itself, Running the Network The network is used and outputs generated. These may be checked against expectations and further balancing done as the network learns and improves.
Applications of ANS Post Office has been using ANS to automate the reading of postcodes An ANS system to predict the stock market NASA have used ANS systems to pilot aircraft
Adv/Disadv of ANS Advantages They can learn and adjust to changing circumstances They have a good success rate at predicting the correct response Disadvantages The difficulty is setting them up and training them, determining why they are giving particular responses

Artificial Intelligence

  • 1.
  • 2.
    Aims of Lesson1 What is Intelligence?? Why is there so much confusion to its meaning
  • 3.
    What is Intelligence????“Intelligence is the name we give to the data processing activity of entities which respond to information with behaviour which appears to be intended to be optional with respect to pre-set goals” “Intelligence involves Knowing and Choosing”
  • 4.
    Intelligence?? Many researchersfeel that intelligence has the following features The Ability to: Learn and adapt from experience. Use knowledge to make decisions. Problem solve. Show Creativity Handle and manipulate language “ Artificial intelligence is concerned with building machines that can act and react appropriately, adapting their response to the demands of situation.”
  • 5.
    Confusion to itsMeaning?? It is a subject which is developing all the time, It is a subject which involves computer scientists, biologists, psychologists etc and they all have a different perspective It is difficult to define human intelligence, let alone artificial intelligence.
  • 6.
    Aims of Lesson2 Test for Identifying Intelligence Early Developments in AI Game Playing
  • 7.
    Testing Intelligence TuringTest Alan Turing, a British mathematician developed a test in 1950 to determine if a program was intelligent: It has the following features: A human tester was connected to 2 terminals and asked a series of questions. One of the terminals had another human making the responses and the other used a computer program to make the responses If the human tester could not distinguish between the human responder and the program response then the software was said to be intelligent.
  • 8.
    Problems with TuringTest It needed a fairly limited problem domain i.e. the area of the knowledge had to be small. Did the program pretend to forget things or make errors in order to fool the human tester?
  • 9.
    Early Developments inAI (1940-65) In the beginning the focus of AI research was on modelling the human brain. Research shifted to using games like noughts and crosses, chess etc to create “AI” systems These games were effective as they had limited problem domains. The games also had a number of rules that were easy to define.
  • 10.
    Aims of Lesson3 Last Lesson Definition of intelligence Why can experts not agree on the definition of AI The Turing Test Early Developments in AI = Game Playing Today’s Lesson Language Processing ELIZA SHRDLU PARRY Chatterbots
  • 11.
    Language Processing (1965-1975)In 1965 Researchers agreed that game playing programs could not pass the Turing test The focus shifted to language processing ELIZA (1966) 1 st language processing program Responded to users inputs by asking questions based on previous responses
  • 12.
    Language Processing (1965-1975)PARRY (1972) Parry modelled a conversation with a paranoid person This seems odd but the program was created by a psychiatrist SHRDLU (1970) The program could interpret verbal commands to move coloured blocks "Move the red block behind the green one" It could understand and carry out the instruction Chatterbots??
  • 13.
    Aims of Lesson4 Last Lesson Definition of intelligence Why can experts not agree on the definition of AI The Turing Test Early Developments in AI = Game Playing Language Processing ELIZA SHRDLU PARRY Chatterbots Today’s Lesson Knowledge representation Semantic Nets Logic programming List Prolog Developments in Hardware AI questions on Forum
  • 14.
    Knowledge Representation Semanticnets Diagrams which show how the information in a system is interlinked They can get very large and out of hand if you try to model a lot of knowledge
  • 15.
    Knowledge Representation LogicProgramming Implemented using declarative languages. Declarative Languages They have no fixed start or end point Consist of collection of facts and rules Use goal directed problem solving Prolog/Lisp
  • 16.
    Knowledge Representation EarlyAI developments were hampered by the programming techniques used human conversations cannot easily be represented by an algorithm A framework for representing knowledge' was a crucial moment for AI semantic nets logic programming
  • 17.
    Aims of Lesson5 Last Lesson Definition of intelligence Why can experts not agree on the definition of AI The Turing Test Early Developments in AI = Game Playing Language Processing ELIZA SHRDLU PARRY Chatterbots Knowledge representation Semantic Nets Logic programming List Prolog Today’s Lesson Developments in Hardware Applications of AI Intro to Natural Language Processing Uses of NLP NLP Steps
  • 18.
    Developments in HardwareFaster processors More memory Increased backing storage capacity Computers being developed with multiple processors (parallel processing where a task can be processed simultaneously on different processors)
  • 19.
    Uses of NaturalLanguage Processing Customer query lines Vue cinema Bank of Scotland O2 Language translation e.g. English to French Speech-driven word processors Command and control systems – giving verbal instructions to save, print etc
  • 20.
    Problems with NLPAmbiguity of words and phrases I saw a man eating fish I saw a man hitting a boy with a stick Changing English language i.e. bouncebackability Accents and dialects i.e. I ken what you are talking about!! Similar sounding words and phrases e.g. furry boots are you fae?? Inconsistencies with grammar
  • 21.
    Aims of Lesson6 Last Lesson Definition of intelligence Why can experts not agree on the definition of AI The Turing Test Early Developments in AI = Game Playing Language Processing ELIZA SHRDLU PARRY Chatterbots Knowledge representation Semantic Nets Logic programming List Prolog Developments in Hardware Applications of AI Intro to Natural Language Processing Today’s Lesson Steps in the NLP process Moral, Legal issues with NLP
  • 22.
    Main Stages inNLP 1. Speech Recognition 2. Natural Language Recognition 3. Natural Language Generation 4. Speech Synthesis
  • 23.
    NLP Process SpeechRecognition Get sound input using microphone, sample and convert to digital data Segment the sound into recognisable sounds (phonemes) Natural Language Recognition Put these sounds together to form words, sentences etc “ Understand" the meaning of these sentences Natural Language Generation Create an appropriate response Speech Synthesis Convert the response into intelligible sound output
  • 24.
    Moral and LegalIssues Moral NLP has benefited blind or disabled users who could not access computers In military applications, it could be used in weapon systems Legal Mistranslations leading to commercial problems Possible wrong diagnosis
  • 25.
    Aims of Lesson7 Last Lesson Definition of intelligence Why can experts not agree on the definition of AI The Turing Test Early Developments in AI = Game Playing Language Processing ELIZA SHRDLU PARRY Chatterbots Knowledge representation Semantic Nets Logic programming List Prolog Developments in Hardware Applications of AI Natural Language Processing Today’s Lesson Vision Systems Applications of Vision Systems Vision System Process
  • 26.
    Vision Systems Theaim is to develop systems which can see, make sense of what they see, and react accordingly. Uses Security systems, recognising faces at airports Inspection of manufactured goods judging quality of production Vision systems on automated cars Visual stock control systems Control of AGVs Artificial Visual Sensing for the blind
  • 27.
    Vision Systems ImageAcquisition Signal Processing Edge detection
  • 28.
    Vision Systems Objectrecognition (Pattern Matching) Its a Tin of Soup! Image Understanding
  • 29.
    Vision System ProcessImage acquisition Capturing the image and converting to digital Signal processing improving the resolution, removing distortion Edge detection Identifying edges of image Object recognition Comparing the objects with known objects. Image understanding How do objects relate to each other? What do they mean?
  • 30.
    Aims of Lesson8 Last Lesson Applications of AI Natural Language Processing Uses Problems Steps in NLP process Moral, Legal issues Vision Systems Uses 5 steps in process Today’s Lesson Expert Systems Examples in everyday life Advantages of Expert Systems Disadvantages of expert systems
  • 31.
    Expert Systems Anexpert system is a computer program which uses a knowledge base of facts and rules populated by a human expert The expert system will give reliable advice on a limited area of expertise, and can interact and explain it reasoning to the user (justification facilities) Examples MYCIN to diagnose blood disorders Legal advice Chemical analysis, DENDRAL was designed to identify unknown substances Car mechanic expert system BABY used to monitor premature babies
  • 32.
    Advantages of ExpertSystems Available to access 24/7 Expert System will never get tired It will never have an “Off day” You can have multiple copies of the expert system Can combine the knowledge of many experts No human emotion involved No possibility of expert system retiring No barrier due to poor communication skills
  • 33.
    Disadvantages of ExpertSystems No common sense is used If the expert system makes an error who is repsonsible? Developer? Company using it? Less trust in expert system?
  • 34.
    Aims of Lesson9 Last Lesson Applications of AI Natural Language Processing Uses Problems Steps in NLP process Moral, Legal issues Vision Systems Uses 5 steps in process Expert Systems Examples in everyday life Advantages of Expert Systems Disadvantages of expert systems Today’s Lesson Expert Systems Parts of an Expert System Expert System Shells Moral/Legal issues
  • 35.
    Parts of anExpert System User Interface This allows the user to input information into the Expert System and get responses in a language that the user understands Knowledge base Consists of all the facts and rules about the topic This information will have been provided by a domain experts Converted into an appropriate form by a knowledge engineer. Inference engine , This searches the knowledge base, comes to conclusions and generates questions to the user
  • 36.
    Expert System ShellsAn expert system shell is an expert system with no knowledge base. It consists only of a user interface and an inference engine. EMYCIN - The interface was so good on this expert system that the company sold this as a shell and the user added the knowledge base
  • 37.
    Legal and MoralIssues Legal If the system is wrong, who do you blame? Human expert, company who bought it, company who created it, the programmer Moral Would you like a machine to diagnose serious health problems Expert system wouldn't make a decision on what is morally correct, it would only use facts and rules It wouldn’t consider feelings Is it right to potentially make human experts redundant
  • 38.
    Aims of Lesson9 Last Lesson Applications of AI Natural Language Processing Uses Problems Steps in NLP process Moral, Legal issues Vision Systems Uses 5 steps in process Expert Systems Examples in everyday life Advantages of Expert Systems Disadvantages of expert systems Parts of an Expert System Expert System Shells Moral/Legal issues Today’s Lesson Intelligent Robots Smart Embedded Technology
  • 39.
    Intelligent Robots Robotscan be considered intelligent when they go beyond simple sensors and feedback (dumb robots), and display some further aspect of human-like behaviour Vision Systems The ability to learn and improve performance Robot that can walk rather than on wheels NLP response Examples Bomb disposal Deep sea exploration and rescue Space exploration Honda’s Asimo
  • 40.
    Practical Problems withIntelligent Robots Provide a power supply? Implement effective 3D vision system? Implement mobility? Providing adequate processing power? Manipulating objects?
  • 41.
    Smart Embedded TechnologySET is where a processor is inside another device, making the device “smart” e.g. it can make decisions and learn from experiences Examples The car that can park itself without you The fridge that knows when food orders are running low and orders in more food Car engine control systems which monitor performance and inform the user when maintenance is required The tv system that works out what you might enjoy watching from past viewing patterns and records it when you are out
  • 42.
    Aims of Lesson10 Last Lesson Applications of AI Natural Language Processing Uses Problems Steps in NLP process Moral, Legal issues Vision Systems Uses 5 steps in process Expert Systems Examples in everyday life Advantages of Expert Systems Disadvantages of expert systems Parts of an Expert System Expert System Shells Moral/Legal issues Intelligent Robots Smart Embedded Technology Today’s Lesson Intro to Artificial Neural Systems
  • 43.
    Artificial Neural SystemsANS is an approach to AI where the developer attempts to model the human brain The human brain is made up of billions of connections called Neutrons. A Neutron will “fire” when it gets enough input from other Neutrons. This firing of Neutrons is how the brain sends signals to the muscles
  • 44.
    ANS Continued AIresearchers have modelled this process using a network of artificial neutrons (perceptron). If you connect a lot of these perceptrons together you can get some fairly complex decisions being made, which might rely on dozens of inputs.
  • 45.
    ANS Continued Simplyspeaking, a neural network is an electronic model of the brain consisting of many interconnected simple processors. Neural networks can either be built in hardware as hard wired circuitry or be implemented in software.
  • 46.
    Aims of Lesson10 Last Lesson Applications of AI Natural Language Processing Uses Problems Steps in NLP process Moral, Legal issues Vision Systems Uses 5 steps in process Expert Systems Examples in everyday life Advantages of Expert Systems Disadvantages of expert systems Parts of an Expert System Expert System Shells Moral/Legal issues Intelligent Robots Smart Embedded Technology Intro to Artificial Neural Systems Today’s Lesson Layers 3 basic stages of ANS creation Applications of ANS Advantages/Disadvantages of ANS
  • 47.
    ANS Layers ANSare made up of layers with input, internal (hidden) and output layers Perceptrons in the input layer react to different stimuli which triggers perceptrons in the hidden layers If the combined output from the hidden layers passes a certain threshold, it will pass an output signal You can therefore make fairly complex decisions based on dozens of separate inputs
  • 48.
    Stages on CreatingANS Initial Setup The network is created and the initial values (weights) are set for the input layer, the hidden layers and the output layers. Training the Network A number of trials are run, to test the accuracy of the original threshold values. The internal weights change, or rebalanced, so that the network gives the correct outputs. These changes may be done manually or the network may rebalance itself, Running the Network The network is used and outputs generated. These may be checked against expectations and further balancing done as the network learns and improves.
  • 49.
    Applications of ANSPost Office has been using ANS to automate the reading of postcodes An ANS system to predict the stock market NASA have used ANS systems to pilot aircraft
  • 50.
    Adv/Disadv of ANSAdvantages They can learn and adjust to changing circumstances They have a good success rate at predicting the correct response Disadvantages The difficulty is setting them up and training them, determining why they are giving particular responses