Artificial Intelligence Methods
Part I – General Introduction – Population-based Search
This coursework is a simple but more challenging extension of coursework 1. The aim of coursework 1 was to familiarize you with the neural network framework adopted, and how to perform basic algorithmic modifications and comparisons.
The aim of this coursework is to push you to design and implement more complex search processes, and to focus on more stringent performance control. In particular, coursework 2 will focus on: population-based search with the incorporation of memetic algorithms.
This coursework will use the same codebase as coursework 1, and you should feel free to proceed from your updated coursework 1 codebase, if you wish.
Coursework 2 will follow the same “item based” structure for your deliverables as in coursework 1, however, your design and implementation challenges will be greater, and your report will also be more stringent in terms of the expected content quality/depth.
As before, it is not necessary for you to modify the neural network framework, however, if you are interested in this aspect, please feel free to do so. Having said this, please keep in mind that the main focus of the coursework should be the architectural search component.
Part II – Coursework Description
As before, the coursework can be broken down into a simple list of items/features that you need to implement. Although the number of items is smaller compared to coursework 1, each item will demand a greater extent of innovation, coding, and testing.
As before, the coursework items are divided across two main categories: (1) search algorithm, and (2) performance evaluation. You can again select your preferred performance measurement, e.g.: (1) higher test accuracy (controlled number of evaluations), (2) smaller number of evaluations (controlled test accuracy), or (3) both higher test accuracy and smaller number of evaluations. For the latter category (performance evaluation), you need to implement a systematic comparison between your modified algorithm from coursework 1 and your proposed algorithm for coursework 2. As before, you will need to run
multiple trials for each condition, and compute some basic statistics to compare the conditions (e.g. mean, standard error, etc.).
Coursework 2 items pertaining to the search algorithm:
Item 1. Implement a population based search
Item 2. Incorporate local learning (memetic algorithms).
Item 3. Include some variant of differential search. Coursework 2 items pertaining to performance evaluation:
Item 1. Implement a clear performance
Item 2. Generate several simple descriptive results (e.g. mean, standard errors, boxplots, ).
Item 3. Run comparison statistics (e.g. unpaired two-samples Wilcoxon test)
Part III – Deliverables and Marking Criteria
The main deliverables for coursework 2 are your code, and a brief report. Your report will follow the same simple structure as the one for coursework 1, where each section corresponds to a different item, and within which you need to briefly explain what you have done. Appendices are allowed, however, very clear references to them must be made from within the main text, and these Appendices should not be seen as a convenient solution/alternative to the page limit. The page limit is again constrained to two pages, so you must deliberate very carefully on how to summarize and portray your work.
As mentioned before, coursework 2 is more demanding than coursework 1 in terms of the expected quality of your solutions. You should no longer be thinking about simple/minimalistic modifications, but rather inventive and well-argued modifications. Apart from that, you will need to code a population-based approach from scratch, however, you are free to choose whatever population- based approach you are more interested in (e.g. genetic algorithms, particle swarm optimization, etc.).
In summary, the main deliverables are: (1) code, (2) two-page report (font: Calibri size 11). More details on how to submit your code and report will become available in Moodle. Two pages is the maximum limit.
Refer to the next page for more details on the marking criteria.
General Assessment Criteria
Code: excellent implementation of all items; concrete evidence of innovative
Report: excellent, well-written report demonstrating extensive understanding and good
Code: good implementation of all items; limited innovation and/or excessive
Report: a comprehensive, well-written report demonstrating thorough understanding and some
Code: simple implementation of items; no
Report: a competent report demonstrating good understanding of the implementation.
Code: very basic implementation of items; some items partially implemented.
Report: an adequate report covering all specified items at a basic level of
Code: missing and incomplete
Report: an inadequate report failing to cover the specified
DescriptionIn this final assignment, the students will demonstrate their ability to apply two ma
Path finding involves finding a path from A to B. Typically we want the path to have certain properties,such as being the shortest or to avoid going t
Develop a program to emulate a purchase transaction at a retail store. Thisprogram will have two classes, a LineItem class and a Transaction class. Th
1 Project 1 Introduction - the SeaPort Project series For this set of projects for the course, we wish to simulate some of the aspects of a number of
1 Project 2 Introduction - the SeaPort Project series For this set of projects for the course, we wish to simulate some of the aspects of a number of