Research Seminars

Research Seminars

The Artificial Intelligence & Robotics Society grants members exclusive access to PhD research seminars - normally only available to academic staff.

Be aware that these sessions sometimes covers advanced topics in mathematics and/or computer science that are not taught in undergraduate level.

Research Seminar 7

2019-2020: Research Seminar 7

Using machine learning to analyse student assessment feedback

Speaker: Zainab Mutlaq-Ibrahim

Date: 04/12/2019

Time: 14:00-15:00

Location: Buckingham Building 0.20

Assessment constitutes a fundamental part of an academic Learning process due to its importance in testing students gaining knowledge and finalizing their grades. This study aims to develop a data mining based framework for analysing students' assessment feedback that will be obtained from social media sites and/or text feedback such as end of unit feedback. The study consists of three stages: The first stage is to build a model that automatically detect the polarity of student feedback using sentiment analysis methods.

The second stage is to build a model that automatically identify and classify issues of assessment. And finally, test the correlation between issue(s) and students' performance.

The research uses Support Vector machine (SVM), Naïve Bayes (NB), decision Tree (DT), and Random Forest (RF) algorithms for text classification to analyse students' feedback of assessment to enhance learning process.

Research Seminar 6

2019-2020: Research Seminar 6

Machine Learning & Artificial Intelligence

Speaker: Fatima Chiroma & Obinwa Ozonze

Date: 04/12/2019

Time: 14:00-15:00

Location: Buckingham Building 0.20

This week plays host to do fantastic speakers. First, Fatima will present a talk on "Detection of Suicidal Twitter Posts". After that, Obinwa will be doing a talk on "Towards Improving the Quality of Health Data with Artificial Intelligence".

Research Seminar 5

2019-2020: Research Seminar 5

Understanding the role of the Care and Health Information Exchange in changing clinical practice: a realist evaluation

Speaker: Elisavet Andrikopoulou and Philip Scott

Date: 6/11/2019

Time: 14:00-15:00

Location: Buckingham Building 0.20

The overall evidence for the impact of electronic information systems on cost, quality and safety of healthcare remains contested. Whilst it seems intuitively obvious that having more data about a patient will improve care, the mechanisms by which information availability is translated into better decision-making are not well understood. Furthermore, there is the risk of data overload creating a negative outcome. There are situations where a key information summary can be more useful than a rich record.

Research Seminar 4

2019-2020: Research Seminar 4

Deep Fuzzy Models

Speaker: Alexander Gegov

Date: 16/10/2019

Time: 14:00-15:00

Location: Buckingham Building 0.20

Alexander Gegov will be presenting a short version of the tutorial on Deep Fuzzy Models that he presented with two international research collaborators at the IEEE International Conference on Fuzzy Systems in June 2019.

The presentation will discuss synergies between Deep Learning and Fuzzy Systems.

You can find more details about this presentation here:

Research Seminar 3

2019-2020: Research Seminar 3

A Systematic Approach to Implementing Artificial Consciousness using Machine Learning and Theory of Evolution

Speaker: Mark Godfrey

Date: 09/10/2019

Time: 14:00-15:00

Location: Buckingham Building 0.20

Research Seminar 2

2019-2020: Research Seminar 2

Enabling local people and groups to support global organisational development

Our organisations emerge from networks of autonomous people engaged in interaction processes (Espejo & Foss, 2018). People, in collectives, use their skills, resources and capabilities to create and produce whatever outcomes they may wish to achieve. Collaboration in these interactions, to a significant degree, depend on processes of self-organization. In general there is no one with authority to tell people what to do and how to interact; they just interact. Often these interactions are inadequate and it is only through learning processes, which depend on cues and signals, that they proceed towards desirable outcomes. To a degree this is the dynamics of organisational development to respond to environmental, social, and economic pressures. Self-organising processes are at the core of their interactions. In today’s world technologies, digital and others, are transforming these interaction processes. New forms of communication and relationships are emerging between people and their environments; these are processes towards the constitution of effective organisational systems (Beer, 1979, 1985), (Espejo & Reyes, 2011). However, these systems are more than the outcome of bottom-up self-organisation; they are also, the outcome of guided self-organisation, which, through policies clarify purposes and help to speed up learning processes by enabling relating fragmented resources. Organisational development and problem solving require of both; bottom-up and top-down interactions. The challenge is working out which interaction strategies are necessary to increase response capacity to make sense of an often overwhelmingly complex surrounding. These are aspects related to Ross Ashby´s law of requisite variety (Ashby, 1964). We learn to manage these interactions often at a high cost to people and organisation; hierarchical structures tend to concentrate responses to environmental challenges at the top of the organisation. On the other hand heterarchical organisations try to distribute response capacity and self-organisation throughout the collective, but often their local response capacity is limited by resources. However, current information and communications technologies are increasing the chances of making this distribution effective.

Speaker: Professor Raul Espejo

Date: 02/10/2019

Time: 14:00-15:00

Location: Richmond Building LT2

Research Seminar 1

2019-2020: Research Seminar 1

On Minimisation of Treewidth and Fill-In

In graph theory, a chord is defined as an edge between a pair of non-adjacent vertices of a cycle. After the addition of a set of chords, a graph becomes triangulated if every cycle has a chord. In a minimal triangulation, the removal of any chord creates a chord-less cycle. First, we review some efficient algorithm for minimal triangulation.

Second, we consider two well-studied problems in graph theory with various real-world applications. Both problems can be solved by finding specific triangulations of the input graph. The Minimum Fill-In problem is the problem of finding a triangulation with the minimum number of chords whereas the Treewidth problem corresponds to the problem of finding a triangulation such that the cliquesize (size of the largest set of pairwise adjacent vertices) is kept to a minimum. It has been proved that neither of the two problems can be solved efficiently. 

We overview the graph classes where the answer to the two problems differ and introduce a class of graphs that contains all the aforementioned classes. We then discuss the closely related TFM (the Treewidth and Fill-in Minimisation) problem where given a graph G, positive integers c and k the output is 'yes' if there exists a triangulation of G with cliquesize of at most k more than optimal, that has at most c many more chords than the minimum number of chords needed. We point out the values of k and c for which the TFM problem has a positive answer in our class of graphs. Finally we study some well-known graph classes where we believe the TFM problem has a positive solution for any c and k.

Speaker: Mani Ghahremani

Date: 25/09/2019

Time: 14:00-15:00

Location: Buckingham Building 0.20

Looking for last year's sessions?

You can find 2018-19's sessions by clicking here.