Hello, if you have any need, please feel free to consult us, this is my wechat: wx91due

ENG 335 and KNE 441

Intelligent Systems

2024

ASSIGNMENT 2

Artificial neural networks for classification of Iris plants

This is a compulsory assessment item. It counts 15% towards the final assessment and contributes to learning outcome ILO6. ILO6 is assessed in this assignment and a mark of 50% is required to achieve this ILO.

Goals:

The Iris plant data set provided below contains 3 classes of 50 elements each, where each class refers to a type of the Iris plant. Each plant is characterised by its sepal length, sepal width, petal length and petal width.

(a) Divide randomly the Iris data set into training and test sets. The test set should have around 50 elements.

(b) Create a three-layer back-propagation neural network and train it to classify Iris plants. Test the network using the test data and determine the recognition error.

(c) Create a single-layer competitive network to perform the same classification task. Test the network using the test data and determine its recognition error.

Submission Requirements:

Each student submits an individual report (you should compare performance of a three-layer back-propagation neural network with a single-layer competitive network and discuss your results) as well as software developed.

Plagiarism:

Each assignment must be entirely your own work. Plagiarism is not tolerated (you will automatically fail the course).

Guidelines:

This assignment should take about 6 hours of work. Remembering that a report is required, you should aim to allocate your efforts in roughly the following proportions:

1. Familiarisation with MATLAB Neural Network Toolbox              10%.

2. Implementation of Graphical User Interface (GUI) with input validation 20%

2. Implementation and Testing (part 1)         30%.

3. Implementation and Testing (part 2)         30%.

4. Assignment Report               10%.

Assignment report should include the following:

1. Introduction.

2. A short description of the domain problem.

3. A short overview of the Neural Network toolbox/library and the system developed

4. Comparison of the performance of a three-layer back-propagation neural network with a single-layer competitive network and discussion of the results.

5. Conclusions.

Submission Requirement

1. A pdf version of the report.

2. A zip file containing all the MATLAB project source code (including the GUI).