This assignment requires students to apply machine learning knowledge developed through lectures and hands-on tutorials to analyse a dataset representing a real-world problem. Students will implement the key stages of the machine learning (ML) workflow—data exploration, preprocessing, model selection, training, evaluation, and interpretation—while analysing, evaluating, and justifying the decisions made at each stage based on evidence from the dataset and concepts covered in the unit. Programming will be done in Python using appropriate machine learning libraries and submitted as an annotated IPython notebook containing executable code, explanations, and the outputs generated from running each cell. The notebook must be runnable in Jupyter Notebook or Google Colab and clearly document the progression of the analysis. In addition, students will submit a concise PowerPoint presentation that explains, in their own words, the key stages of the ML workflow, the rationale for major decisions, and visualisations of the results. The final component is a short viva conducted in Week 6 tutorials, during which students will orally explain and defend their analysis using the PowerPoint slides.

The marking criteria are structured around the three submitted components of the assessment. The annotated IPython notebook is assessed on the correctness and completeness of the machine learning implementation in Python, the quality of embedded explanations and annotations, the alignment between code, outputs, and interpretations, and the justification of analytical decisions supported by evidence from the dataset. The PowerPoint presentation is assessed on the clarity and coherence with which students summarise the machine learning workflow, explain key decisions in their own words, and use appropriate data visualisations to communicate results and limitations. The viva is assessed on students’ ability to orally explain, justify, and defend their analysis, respond to questions about their modelling choices and results, and demonstrate conceptual understanding of the work they have submitted.
        
The marking criteria will be further expanded on the unit Moodle. Please ensure to read through the marking criteria carefully before submitting your work.

AI ASSESSMENT SCALE - AI PLANNING

You may use Al for planning, idea development, and research. Your final submission should show how you have developed and refined these ideas.

Note: This unit involves programming. For programming assignments you can get examples and ideas from AI, but you must not include code in your program that was generated by AI. You must write the code in your application yourself.  If asked, you must be able to explain any of the code submitted and you must be capable of writing similar code under invigilated, test conditions if required to do so.  In addition, you must only use the language features and techniques covered in the unit.

This assignment assesses students’ conceptual understanding of heuristic and metaheuristic search techniques through the analysis of a practical optimisation or search problem. Students will be provided with a real-world problem scenario characterised by a large or complex search space, where exhaustive or optimal search methods are impractical. Drawing on content covered in related lectures and tutorials, students are required to analyse the problem, explain why classical search techniques such as breadth-first or depth-first search may be insufficient, and justify the use of one or more heuristic or metaheuristic approaches, such as Genetic Algorithms or Particle Swarm Optimisation.

Students are encouraged to support their reasoning by referencing relevant publications from the literature (e.g. textbooks, review papers, or applied studies) to justify modelling choices and trade-offs, with all sources clearly and correctly cited. Students must articulate how the chosen method(s) would represent solutions, guide the search process, and balance considerations such as solution quality, computational cost, and time constraints. Rather than describing algorithms in isolation, the focus is on reasoning about algorithm–problem fit, limitations, and practical considerations. The submission is a concise written analysis of no more than 1000 words (including references).

There are several main marking criteria, including "Problem Characterisation", "Analysis of Classical Search Limitations", "Heuristic or Metaheuristic Reasoning", "Trade-offs, Risks, and Limitations", and "Reflection on Practical Use".

The marking criteria will be provided on the unit Moodle. Please ensure to read through the marking criteria carefully before submitting your work.

AI ASSESSMENT SCALE - AI PLANNING

You may use Al for planning, idea development, and research. Your final submission should show how you have developed and refined these ideas.

This assignment requires students to work in groups to apply deep learning techniques to a real-world problem, using Python to implement and evaluate their solution, while demonstrating individual understanding and contribution. As a group, students will preprocess a dataset, adapt and fine-tune one or more pre-trained deep learning models, and design and evaluate the training process. Students will analyse model performance, interpret results, and identify limitations in relation to the problem context.

The group submission includes an annotated IPython notebook, which documents the implementation, explanations, and outputs of the analysis, along with a written report and a recorded presentation that summarise the approach, results, and key design decisions. In addition, each student must submit an individual reflection that clearly describes their personal contributions to the project, including the aspects of the implementation or analysis they were responsible for, challenges encountered, and lessons learned. Assessment will consider both the quality of the group outcomes and each student’s demonstrated understanding and ownership of the work.

The marking criteria cover both group and individual components. The annotated IPython notebook is assessed on the correctness of the Python implementation, data preparation, and the alignment between code, explanations, and outputs. Model selection and adaptation evaluates the suitability of the chosen pre-trained model(s) and how effectively they are applied to the task. Model training and evaluation focuses on the training design, use of appropriate performance metrics, and interpretation of results. The written report and recorded presentation are assessed on clarity, organisation, and effective communication of key decisions and results. The individual reflection assesses each student’s contribution, understanding, challenges encountered, and insights gained from the group project.

The marking criteria will be provided on the unit Moodle. Please ensure to read through the marking criteria carefully before submitting your work.

AI ASSESSMENT SCALE - AI PLANNING

You may use Al for planning, idea development, and research. Your final submission should show how you have developed and refined these ideas.

Note: This unit involves programming. For programming assignments you can get examples and ideas from AI, but you must not include code in your program that was generated by AI. You must write the code in your application yourself.  If asked, you must be able to explain any of the code submitted and you must be capable of writing similar code under invigilated, test conditions if required to do so.  In addition, you must only use the language features and techniques covered in the unit.