70061 Research Tutorials in AI and Machine Learning

 

Overview

This is an interactive module that aims to develop students' ability to critique, summarise and present scientific literature. The focus will be on learning the principles of high-quality academic research in artificial intelligence (AI) and machine learning including theory, novel algorithmic work, empirical research, as well as literature on AI ethics.

The teaching will be carried out through weekly reading group tutorials in which students present and critique a published paper, under the close supervision of the module leaders.  Additionally, after the sessions, students will write reports summarising the paper and major points of the discussion.

At the end of the module, students will be able to evaluate, appraise, and critically review published research work, as well as develop a deeper understanding of AI sub-domains.

Learning outcomes

Upon completion of this module students should be able to:

• Extract and evaluate key information from state-of-the-art work in AI conference proceedings and high impact journal publications covering theoretical, algorithmic, empirical, and applied research.

• Assess how and why certain research work transformed the understanding of their respective AI sub-domain (e.g. computer vision, NLP) and why they may have transformative impact on other areas of science.

• Identify strengths and limitations of publications, by considering clarity of hypothesis, datasets used, claims for novelty, importance, and value; algorithmic or empirical contribution, as well as reproducibility of the work.

• Distil complex research concepts into clear, structured summaries; demonstrate ability to lead group discussions and to communicate effectively and concisely with peers across multidisciplinary topics.

• Produce high quality technical reports aimed at audience ranging from domain experts to novice readers. Address individual feedback aimed to improving writing style, clarity, succinctness, and good academic practice.

Teaching methods

A weekly session will take place in the first and second term through a flipped-classroom approach. The sessions will be interactive, and the student leading the discussion will produce a presentation of key insights from the paper and lead the discussion on the algorithmic and/or empirical aspects of the paper.

Assessment

Formal assessment

This module is assessed through the delivery of a presentation on the paper chosen by the student, in addition to the production of a written report that summarises both the paper and the discussion that took place during the tutorial session. The assessment will consider students' critical analysis of the paper, demonstrated knowledge of the paper content and relevant research landscape, as well as the student’s skills in answering questions. Additionally, every student will be expected to actively contribute to the discussion in other students’ presentations and provide peer feedback.

Peer feedback

We encourage a culture of collaborative research where students can freely discuss new innovative ideas in an open way and provide constructive criticism respectfully. Therefore, after each presentation, each student in the audience will be asked to provide feedback on one aspect they liked the most about the presentation, and also a suggestion of one area of improvement.

Module leader

Dr Ahmed Fetit