Reference no: EM132476941
COMP7212 - Artificial Intelligence - TOI-OHOMAI Institute of Technology
Learning outcome 1: Research and analyse the general nature of Artificial Intelligence and the problems it solves
Learning outcome 2: Apply and evaluate artificial intelligence techniques for solving a variety of real world problems
Assessment -1: Section-1
AStar Search
Problem 1:
You are to edit and extend the supplied code to accomplish the following tasks. You have a Sectionial code listing for an AStar search of a matrix structure. The code is incomplete and requires your input to make it work. You will need to perform basic research on AStar algorithms to define their functionality and operational characteristics, then use that knowledge to enhance the supplied code.
Initialise the two arrays for holding node data Enter the node neighbour finding code
Create a searchable Matrix:
» Create a searchable Matrix (Grid/Graph) "maze" that is 30 cells wide and 30 rows deep
» The matrix must include a minimum of 10 "walls" ( a wall is a node that the program must recognise as an impassable barrier. A wall must be at least 7 cells/nodes long as a contiguous string.
» The walls must include at least 3 vertical walls, 3 horizontal walls, 4 angled/curved, shaped walls.
» The walls must be placed in such a manner as to create a complex pathway from the starting node to the end node. The walls must obstruct the path of the search, there must not exist any straight path or an easy to navigate path between starting to ending nodes.
» The start node must be located somewhere on the upper quarter lefthand side of the matrix, the end node must be located somewhere on the far righthand side lower half of the matrix
Extend and display the code as below:
» Print the starting and ending node coordinates on screen, once
» Print the recurring current node position on screen for each program loop
» Display each of the child.g, child.h, child.f node values as the program progresses. For each loop display the updated values assigned by the program code to the child.g, child.h, child.f values, create the code/maths sequences necessary to support these displays.
o The child.h position is derived from two different positions as per the default calculation. You are to create code to display each of these positions as a separate numeric value for display on screen.
o Your code is to display the running values of the mathematical relationship between the G,H ,F values
» Display a message on screen each time the program finds a cell that is NOT walkable/navigable (a wall cell)
» Print a running count of the number of times the program runs to discover the end node
Written Section:
You are to accurately describe in a written report each of the following concepts of an AStar search and what their function is as Section of the program
» start_node
» end_node
» heuristic
» g value, f value, h value
» open_list
» closed_list
» current_node
» open_list.append(start node)
» tuples
» visited_nodes
» goal_found
» walkable
Q1. Describe how the program recognises the difference between a valid move ( an empty cell) and an invalid move (a blockage/wall)
Q2. Describe how the program finds and tests each of its neighbour nodes in the structure.
Q3. Describe how and why values from the Open-list will interact with the Closed-list
Q4. Describe the role of the Pythagoras maths used in this code. Define what it actually achieves
Q5. Describe what the Manhattan distance is in relation to an AStar search
Assessment -1: Section-2
Neural Network Text Identification TASK 2:
This task will involve you in creating seven new numeric pattern entries for a neural network program to read/recognise (the same number created seven times). The pattern will be created from a notepad/excel type document. Each student will be assigned a different number pattern to create for use within the neural network program.
Constraints:
The numeric file will be created using numbers between the range of 0 to 255.
Number Zero (0) is used as the numeric blank background, most of the numeric image will be zero's
Numbers from 1 to 255 relate to colour intensity appearing in Grey scale. 1 is almost white and 255 is total black. You are to create the generic shape of the number you have been assigned by writing number values in the 100-255 (realistically higher values will apply for this exercise) range as entries in a csv formatted 28*28 array. REMEMBER the array includes one additional row at the very beginning that has a single digit in it followed by the comma character for labelling the data pattern that follows, technically the complete array is 29 rows long.
Once the array is created as a csv, you must turn that data into a single long string of comma separated values (remove the line-breaks. The number image must appear as a single line of csv text in the excel file for the neural network to read).
When you have created seven different versions of the number copy/insert them into the excel master file.
You are to locate your copy of the numbers in a random sequence inside the master excel file. Do not insert them one after the other inside the excel file.
PYTHON:
Use Python to :
» Select and display the correct array entry of your own number as a graphic image. Create multiple copies of the python code to select each version of your number , or edit the code to display them all inline serially.
» Use python to read the master file and extract the record sequence containing one of your own numbers training the neural network to obtain a confidence factor.
Written Report:
You are to create a written report defining what Neural Networks are and how they operate. Your report must include and clarify the following aspects as well as other information you find during your research.
» Hidden layers
» Output layer
» Input layer
» Weightings
» Threshold
» Labelled dataset
» Classification
» Correlation between label and dataset
» Supervised learning
» Unsupervised learning
» Clustering
Attachment:- Artificial Intelligence.rar