IPT - A Virtual Approach IPT A Virtual Approach by Peter Whitehouse
Quick Links:
 
 
back
Information and Intelligent Systems Social and Ethical Implications Human Computer Interaction Software and Systems Engineering eXercise Files Course Outline and Assessment A-Z of Geeky Acronyms Terrace Work Program 2004 Sillybus FAQ = Frequently Asked Questions Help
 
 

Artificial Intelligence

eXercise #2


Artificial Neural Networks (ANNS)

  1. COUNTER NETWORK: The first 8 counting numbers are represented in binary as follows:

    Binary Decimal   Binary Decimal
    0000 0 0101 5
    0001 1 0110 6
    0010 2 0111 7
    0011 3 1000 8
    0100 4  
    Create a neural network (call it COUNT.NET on your H: drive) that will accept the binary representation and correctly output the decimal equivalent. Train your neural network and demonstrate it's effectiveness to your teacher.


  2. The TC network : a simple Optical Character Recognition Neural Network

    Consider the following optical characters as all of the regular orientations of the characters 't' and 'c'.

    1 1 1
    0 1 0
    0 1 0
    0 0 1
    1 1 1
    0 0 1
    0 1 0
    0 1 0
    1 1 1
    1 0 0
    1 1 1
    1 0 0
    1 1 1
    1 0 0
    1 1 1
    1 1 1
    1 0 1
    1 0 1
    1 1 1
    0 0 1
    1 1 1
    1 0 1
    1 0 1
    1 1 1

    We could specify a rule for each of the orientations of each letter as follows:

    IF top left cell is on 
    and top centre is on
    and top right is on
    and middle left is off
    and middle centre is on
    and middle right is off
    and bottom left is off
    and bottom centre is on
    and bottom right is off
    THEN the letter is a 'T'.
    INPUTS (1 = On, 0 = Off)
    OUTPUT
    Top Left Top Middle Top Right Middle Left Middle Middle Middle Right Bottom Left Bottom Middle Bottom Right Letter
    1 1 1 0 1 0 0 1 0 10 = T
    1 1 1 1 0 0 1 1 1 20 = C

    (note NSHELL can only currently deal with NUMERIC DATA)

    This system has 9 INPUTS, 1 OUTPUT, 8 RULES (the total number of rows in the matrix).

    CREATE a Neural Network according the specifications above. SAVE the network under the name VISION.NET in your home directory.
    TRAIN the network - this may take a number of minutes (aim to get the minimum error (it should be possible to get 0 error).
    RUN the network using some of the KNOWN data, to see if it classifies them correctly.
    TEST the network on some unknown data like the following:

    1 1 1
    1 0 1
    1 1 1

    ..this should be more like a 'c' than a 't'

    MAINTAIN the network by training it to recognise T, C and L
    Add to its training with letters H, I and X

  3. Design and create a Neural Network to deal with XOR (the truth table follows)

    Input 1 Input 2 Output
    true
    true
    false
    true
    false
    true
    false
    true
    true
    false
    false
    false

    NOTE: the system has 2 inputs, 1 output and 4 rules. Allow the system plenty of time to train, then test it correctly classifies test data.

wonko@wonko.info
©Copyright t 1992..2017+. Edition 25.150117
wonkosite
Creative Commons License
This work is licensed under a
Creative Commons Attribution-NonCommercial-ShareAlike 2.1 Australia License
.