深度学习AI代写｜COM3025 Deep Learning and Advanced AI Coursework Academic Year 2022/23, Semester 2

Overview

The overall Units of Assessment (UoA) for the module are as below:

Theory Test (15%): This will take place in week 9, during the lab hours, through SurreyLearn online test platform. The test will be on the content covered in lab notes/tasks and lecture notes.

Coursework Project (65%): Students will choose one of the projects proposed in this guideline. Students will form project groups with 3-5 members to undertake the project selected by the group. Throughout the course, students are required to read a list of state of art literature, and understand the key techniques and concepts in these readings. A poster and the developed program code should be submitted to Surreylearn in week 12, 15th May 2023, Monday, at 4pm. Each project group will present their work with demonstration of the developed system, followed by a viva. This will take place after the project submission.

Feedback will be given within 15 working days.

Important dates:

Coursework Project Assessment

Coursework poster, code and any supplementary materials should be submitted to SurreyLearn.

There are many useful guidelines on how to design an academic poster.1

The poster can be structured based on the following marking criteria. The poster, developed source code, data (if not available online) should be submitted together with any supplementary materials that demonstrate the quality of the project, which could include further analysis, study,experiments, results etc.

Viva

Students will present their posters and the developed systems during the viva which involves questions and discussions. Questions on all subject areas covered in the module as well as the readings from the literature could be asked.

Marking criteria for the poster and viva

Proposed coursework ideas

The followings are suggested ideas for a coursework project:

1 Skin Cancer Detection and Interpretation

The aim of this project is an opportunity for you to apply what you have learned in class to a real world problem, which is yet to be solved. Much effort in this area have been carried out as seen in publications including data and software. Up to now, there are still live challenges for this particular problem.

There are multiple elements in the research and development that you can consider in this project:

There is a need to clean and organise the data appropriately, partly due to the fact that there are many duplications from multiple data sources; also, it is important to perform data analysis and prepare the data for a holistic AI algorithm development. Apart from the data in the given references, you can also collect further data, so that the overall collection is sufficiently large enough with appropriate distribution of various skin conditions alongside normal ones, for training, validation and testing.

a.Various visualisation techniques can be developed to allow explainability of the software as well as illustrating in-depth clinical information for a better user experience (e.g. aiding doctors/nurses/clinicians in the clinical pathway for more precise diagnosis and better patience’s outcome.)

b.Dive deeper in this problem, to see if AI algorithms can understand similar patterns as what human experts have discovered. Check out here2 .

The following are some references on skin cancer datasets as well as some overview of AI algorithms in primary care setting. These could be a starting point to embark your own research and development. Broader literature review could be carried out.

a.Characteristics of publicly available skin cancer image datasets: a systematic review 3 , 4 , 5

b.Artificial intelligence and machine learning algorithms for early detection of skin cancer in community and primary care settings: a systematic review6 , 7

You may find many other resources, data, code, developed by the research community as well as those through various skin condition detection challenges/competitions.

Any project developed on top of existing work, should carefully cite the source of prior work, and specify your own and additional contribution in the project.