辅导 CMT316 Applications of Machine Learning: Natural Language Processing and Computer Vision Coursewo

Assessment Proforma 2024-25

Key Information

The Assessment Calendar can be found under ‘Assessment & Feedback’ in the COMSC-ORG-SCHOOL organisation on Learning Central. This is the single point of truth for (a) the hand out date and time, (b) the hand in date and time, and (c) the feedback return date for all assessments.

Learning Outcomes

The learning outcomes for this assessment are as follows:

1.  Implement and evaluate machine learning methods to solve a given task

2.  Explain the fundamental principles underlying common machine learning methods

3.  Choose an appropriate machine learning method and data pre-processing strategy to address the needs of a given application setting

4.  Reflect on the importance of data representation for the success of machine learning methods

5.  Critically appraise the ethical implications and societal risks associated with the deployment of machine learning methods

6.  Explain the nature, strengths and limitations of an implemented machine learning technique to an audience of non-specialists

7.  Explain the fundamentals and modern principles of natural language processing or computer vision

Submission Instructions

The coversheet can be found under ‘Assessment & Feedback’ in the COMSC-ORG- SCHOOL organisation on Learning Central.

All files should be submitted via Learning Central.  The submission page can be found under ‘Assessment & Feedback’ in the CMT316 module on Learning Central.  Your submission should consist of multiple files:

This coursework consists of a group project divided into two parts with different weights:

-    Part (1) consists of a group report on a specific machine learning project. The final deliverable consists of a single PDF file and a zip file with the code. The deliverable includes a zip file with the code, and a written summary (up to 4500 words) describing solutions, design choices, evaluation and a reflection on the main challenges faced during development and insights gained throughout the process.

-    Part  (2)  consists  of  an individual reflective  essay (up to  1500  words)  where students reflect on the main insights gained as part of the group project. Cover sheet should also be submitted in this part.

Part 1: The group report should be submitted in learning central (group assignment) by a nominated team member as a single PDF document and a zip file. Prior to handing in, make sure all documentation has been collected. Additional supporting material, such as sources or data may also be submitted if appropriate along with the code zip file. Any code submitted will be run in Python 3 and must be submitted as stipulated in the instructions. All team members must have seen and agreed to the final version of the submission. Make sure the report clearly mentions your group number, a list of all members of the group (with full name and student ID as on learning central), the project title, and the name of the supervisor on the title page of your report.

Part 2: Individual submission for the cover sheet and individual reflective essay.

Any deviation from the submission instructions above (including the number and types of files submitted) may result in a mark of zero for the assessment or question part.

If you are unable to submit your work due to technical difficulties, please submit your work viae-mailto comsc-submissions@cardiff.ac.ukand notify the module leader.

Assessment Description

In this coursework, students demonstrate their familiarity with the topics covered in the module via a group project.

Marks will be awarded to the individual student based on the quality of the group report and their contribution and the  individual  report. All  students should  contribute to the group projects - extenuating circumstances submitted for the spring term project period will be considered pro-rata for the contribution and for an extension on the individual essay.

Part 1: Group report

In Part 1, students will be allocated in groups to design a machine learning project in one

specific topic. The list of all topics along with their descriptions is available in the following link: https://docs.google.com/document/d/1P8jc81L_HW3DDdZaIMfMcekrPeDYuBm- 2qTC7knbKV8/edit?usp=sharing

Each group will be assigned a specific dataset and a supervisor. The task of each group consists of developing a whole machine learning pipeline that attempts to solve the task. The usage of neural networks as methods/baselines is not mandatory but will be positively assessed; the non-usage of neural methods should be properly justified.

Throughout the course the groups should present their progress to their supervisor each session. Finally, the group will write a report summarizing the steps followed and the main insights gained as part of the process.

As part of the group decisions, each student will be allocated to one of the following tasks:

-    Descriptive analysis of the dataset + Error analysis

-    Preprocessing + Literature review

-    Implementation + Results

Each of these tasks will typically have two students involved (except in exceptional cases when this is not possible), who will work together in the specific task and as part of the group. The structure of the report will be decided by the group members. In the following link, students can find some guidelines to write the  report,  including  some of the  common sections that groups may want to include in their report:

https://docs.google.com/document/d/1ku-K6mBH8-

Wdfy_Dz_gvpReDv6knWCFxq4rsMDCrPHY/edit?usp=sharing

Note: These are just guidelines and students are not forced to follow this structure. New sections may be added or adjusted if necessary.

Each student will also be involved in all group activities/tasks and will be responsible for the well functioning and coordination of the team members.

Deliverables

The deliverables for this part include a report of no more than 4500 words and a zip file with the Python code. The code should contain three specific parts:

(1) Code to get the statistics used to complement the descriptive analysis of the dataset.

(2) Code to train one of the best performing models in the training set and evaluate it in the test set. This code should also include all steps for preprocessing the original dataset, if it were necessary.

(3) A README file explaining how to run the code for each of the two parts.

The code will not be marked separately and will only be used as a complement to assess specific parts of the report.

Part 2: Individual reflection essay

In Part 2, students are asked to write a reflective essay about their group projects. The individual essay must discuss your contribution to the group report and to the overall group work. You must show that you contributed to the group work, which will be determined via the individual report and the contribution monitoring, conducted by the supervisor, if it were necessary. Explain what tasks you have performed and provide evidence of your work (you may refer to the group report for the actual work/results). Discuss how you approached these tasks and how you interacted with other members, both in sharing your results and in organising the team's activities. Consider how well your existing skills were utilised and what new skills you have learned. Then reflect on your overall performance and role in the team and suggest what went well and what changes you will be making to improve (1) your performance in particular, and (2) the performance and results of methods and analyses performed as part of the project. You may also reflect on how your perspective and approach changed over time and adapted to improve your work.

Note: Please indicate the information about your group (group number, project name) in a visible place at the top of your essay.

The individual essay must have no more than 1500 words. It does not have to be exhaustive, but should contain good examples of what you have done and discuss key aspects. This part weighs 25% of the total marks.

Assessment Criteria

Criteria for each individual part is provided separately.  The final mark will be obtained from a weighted sum of the two parts: Part 1 - 75%; Part 2 - 25%.

The final mark for Part 1 (75% of the total marks) will result from the following items:

-    Descriptive analysis of the dataset + Error analysis (15%)

-    Preprocessing + Literature review (15%)

-    Implementation + Results (15%)

-    Student’s own allocated task from the three above (15%)

-    Group report as a whole, including its coherence and structure (15%)

Note: In addition to the specific individual task assigned, in some cases marks might be weighted by the individual contribution in the project. This would be based on collected evidence.

All main criteria carry equal weight as  indicated above for your total  mark and will  be evaluated on the following scale:

T1-1. Descriptive analysis of the dataset (7.5%)

T1-2. Result analysis (7.5%)

T2-1. Preprocessing (7.5%)

T2-2. Literature review (7.5%)