2018-2019 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.

These seminars occurred in the 2018/19 academic year.

To view seminars for the current academic year head over to our research seminars page.

Research Seminar 13

2018-2019: Research Seminar 13

Intelligent organization based on viable system model and cybernetics

This presentation aims to revisit cybernetics; synergize Ross Ashby’s Law of requisite variety and Viable System Model (VSM) in order to show its relevance as a theoretical and practical framework in a complex engineering organization.

The application of this model under study can activate major potential for improvement of this key organization(CSTP-NASRDA) by improving communication, control, enhance decision-making within the organs of the organization and overall, making them viable systems within a bigger system.

Speaker: Akinola Kila - a PhD student from the School of Computing

Date: 06/03/2019

Research Seminar 12

2018-2019: Research Seminar 12

Understanding and Generating Visual Media using Machine Learning.

In recent years, we have seen a significant trend that computer vision, which traditionally focuses on image analysis and recognition, and computer graphics where the major interest is to produce improved imagery, are merging. The boundary between them becomes increasingly vague and both benefit significantly from recent development in machine learning. In this talk, I will give an overview of our recent research in this joint area which aims to develop automatic methods to understand and synthesise visual media including images and 3D shapes. In particular, we focus on tasks which are traditionally challenging, including image emotion analysis, image style transfer, as well as 3D shape deformation analysis and transfer.

Speaker: Dr Yukun Lai is a Reader in the Visual Computing group, School of Computer Science and Informatics, Cardiff University, UK.

Date: 20/02/2019

Research Seminar 11

2018-2019: Research Seminar 11

Digital health: A transformational tool for maternal and child healthcare delivery in Nigeria

Digital health is a potent tool for any ambitious, sustainable and scalable health system. There has been widespread adoption of electronic medical records (EMR) in most high-income countries (HICs), but this is not the case in many developing nations. So far, only a few of the health institutions in Nigeria have implemented any form of EMR system especially for the domain of maternal and child health (MCH).

Speaker: Taiwo Adedeji, School of Computing

Date: 13/02/2019

Research Seminar 10

2018-2019: Research Seminar 10

Private Cloud Computing Adoption and Decision Making using Knowledge Based System

The adoption of Cloud Computing (CC) and its effect on cost, technology, location, and the business sector has been increasingly researched. Two crucial challenges with the adoption of CC are the choices of deployment and service models. A comprehensive review of the literature revealed a lack of information addressing these models and the purpose of this study is to review the factors influencing, in particular, the adoption of private cloud computing in an organization. This paper presents a different classification of private cloud computing and also identifies the importance of intranet network as the bases for private CC implementation. Finally, the review suggests some key adoption considerations and related factors, as well as a future analysis of the impact of such factors on private cloud computing.

Speaker: Daniel Olabanji, School of Computing

Date: 06/02/2019

Research Seminar 9

2018-2019: Research Seminar 9

Personalised Context Aware Suggestions

The TREC Contextual Suggestion track (2013-2015) addressed the problem of suggesting contextually relevant attractions to a user visiting a new city based on the user's preferences. In this talk I will reframe the problem of representing and using context and will introduce our two past approaches to capturing the user's profile to enable a system to provide more accurate and relevant recommendations. The results of our participation at TREC showed that our system not only significantly outperforms the baselines, but also achieved the best results in nearly all test contexts. Finally, I will show how this work has been extended to recommending mobile phone apps based on the user current context.

Speaker: Prof Fabio Crestani from Universita della Svizzera Italiana, Switzerland

Date: 30/01/2019

Research Seminar 8

2018-2019: Research Seminar 8

Intelligent Cyber Physical Systems

In this talk, Bo Wei will introduce his research on intelligent cyber physical systems, which includes two main areas: real-time recognition on resource-constrained devices and device-free context awareness using wireless signal.

Cyber physical systems (CPS) have a wide range of applications in natural and urban environment monitoring. Real-time recognition on resource-constrained devices is critically important in CPS because wireless transmission costs several orders of magnitude more energy than computation. The main challenges of in- network classification in CPS include effective feature selection, intensive computation requirement and high noise levels. We propose several methods to address these challenges for various applications, such as acoustic classification, face recognition and indoor localisation.

Context awareness is an important component of many pervasive computing applications. Device-free context awareness has the advantage that it does not have the privacy concern of using cameras and the subjects do not have to carry a device on them. We take advantage of wireless signal for device-free localisation, activity recognition, and identity recognition.

Speaker: Bo Wei from Northumbria University, UK.

Date: 05/12/2018

Research Seminar 7

2018-2019: Research Seminar 7

The quantum information supremacy of linear optics quantum interference

Multiphoton quantum interference underpins fundamental tests of quantum mechanics and quantum technologies, including applications in quantum computing, quantum sensing and quantum communication. Standard quantum information processing schemes rely on the challenging need of generating a large number of identical photons. In this talk, we show how the difference in the photonic spectral properties, instead of being a drawback to overcome in experimental realisations, can be exploited as a remarkable quantum resource. Interestingly, we demonstrate how harnessing the full multiphoton quantum information stored in the photonic spectra by frequency and time resolved correlation measurements in linear interferometers enables the characterisation of multiphoton networks and states, produces a wide variety of multipartite entanglement, and scales-up experimental demonstrations of boson sampling quantum computational supremacy.  These results are therefore of profound interest for future applications of universal spectrally resolved linear optics across fundamental science and quantum technologies with photons with experimentally different spectral properties.

Speaker: Vincenzo Tamma, School of Mathematics and Physics

Date: 28/11/2018

Research Seminar 6

2018-2019: Research Seminar 6

Robotic interaction control and human-robot collaboration

The past decades have seen robots move out from industrial floors to daily lives. They are no longer confined to closed environments but now co-exist and even collaborate with humans in unknown dynamic environments. In this regard, we must carefully deal with the interaction between a robot and its environment, including humans. In this talk, I will introduce some previous and ongoing works on robotic interaction control and human- robot collaboration. Two research questions will be addressed: (i) how to design adaptive or learning strategies for a robot to interact with its environment including humans? (ii) how to best use the complementary strengths of the human and robot in a specific task? These research problems will be elaborated with target applications, including surface exploration, human-robot collaborative manipulation, robotic rehabilitation, etc. Future works and promising research directions will be then outlined.

Speaker: Dr Li is a Lecturer in Control Engineering with the Department of Engineering and Design, University of Sussex, UK.

Date: 21/11/2018

Research Seminar 5

2018-2019: Research Seminar 5

Anomaly Detection - How to Detect What We Don't Know About?

With processing power, data storage units and sensors getting cheaper and cheaper, industrial data can be gathered in greater quantities than ever before in sectors such as aviation, marine and manufacturing. There might be legal requirements to check for safety incidents, or that regulations are being complied with. Additionally, the data may contain valuable information on the presence of faults (known or unknown) or general performance degradation. Industrial sectors have responded differently to this challenge and require more automated methods to successfully extract the information required in a multitude of different environments. Anomaly detection methods have been proposed to tackle some of these challenges and in this seminar, 3 separate case studies in the aviation, the marine and the manufacturing sectors are discussed and results are presented from the anomaly detection methods applied in each case.

Speaker: Dr Edward Smart, a Senior Research Fellow in IIR/SENE

Date: 14/11/2018

Research Seminar 4

2018-2019: Research Seminar 4

Smart Robot Systems for Firefighting Tasks

This talk will first introduce the history of robotics and its classifications. In particular HAM (Human Adaptive Mechatronics) developed from one of my network projects is introduced and discussed. Then the challenges and future researches of robotics in disaster rescues (in particular fire-fighting) are revealed. It will address the part of the above challenges and will provide a review on the robot- assisted firefighting systems with interdisciplinary perspectives to identify the needs, requirements, challenges as well as future trends to facilitate smart and efficient operations. We will consider information acquisition (sensing and visioning technology), transmission (ultra-remote signal transmission) and processing (multi-sensor fusion technology), instrumentation (actuating technology, robotics thermal protection technology), control (multi-degree of freedom mobile and operating robot control methods, obstacle avoidance and sweeping, Decision Support Systems (DSS)) and communication (Human-Robot (H2R) interaction systems, Machine-to-Machine (M2M)). This talk will also introduce the basic requirements to design and develop a hybrid mobile robot and report the recent progress conducted by the EU funded SMOOTH project. Subsequently, prevailing firefighting robotic platforms in literature as well as in practices are elaborately scrutinized and discussed, followed by investigation of localization and navigation support methods.

Speaker: Hongnian Yu from the University of Bournemouth, UK.

Date: 07/11/2018

Research Seminar 3

2018-2019: Research Seminar 3

Synthesizing multi-character interactions using data-driven approaches.

Character animation production pipeline still requires a lot of manual work from artists nowadays. Researchers in the Computer Graphics (CG) community have been trying to develop new approaches for the professionals in the entertainment industry such that they can focus their effort on creativity rather than tedious manual, repetitive tasks. With the advancement of motion capture technology and machine learning techniques, it is possible to develop data-driven approaches to assist artists in daily routine work. My research interests focus on the areas of CG and Computer Vision (CV). In particular, I am interested in applying state-of-the-art machine learning techniques demonstrated in the CV community in solving research problems in CG. In this talk, I will present two recent CG projects.

In the first project, an intuitive way for controlling characters in crowd simulation using multi- touch devices is proposed. We propose a data-driven gesture-based crowd control system, in which the control scheme is learned from example gestures provided by different users. In particular, we build a database with pairwise samples of gestures and crowd motions. To effectively generalize the gesture style of different users, we propose a set of gesture features and crowd motion features that are extracted from a Gaussian mixture model. Given a run-time gesture, our system extracts the K nearest gestures from the database and interpolates the corresponding crowd motions in order to generate the run-time control. Our system is accurate and efficient, making it suitable for real-time applications such as real-time strategy games and interactive animation controls.

In the second project, we introduce a data-driven method to generate a large number of plausible, closely interacting 3D human pose-pairs, for a given motion category, e.g., wrestling or salsa dance. With much difficulty in acquiring close interactions using 3D sensors, our approach utilizes abundant existing video data which cover many human activities. Instead of treating the data generation problem as one of reconstruction, either through 3D acquisition or direct 2D-to-3D data lifting from video annotations, we present a solution based on Markov Chain Monte Carlo (MCMC) sampling. With a focus on efficient sampling over the space of close interactions, rather than pose spaces, we develop a novel representation called interaction coordinates (IC) to encode both poses and their interactions in an integrated manner. Plausibility of a 3D pose-pair is then defined based on the ICs and with respect to the annotated 2D pose-pairs from video. We show that our sampling-based approach is able to efficiently synthesize a large volume of plausible, closely interacting 3D pose-pairs which provide a good coverage of the input 2D pose-pairs.

Speaker: Edmond Shu-lim Ho from Northumbria University, UK.

Date: 24/10/2018

Research Seminar 2

2018-2019: Research Seminar 2

Performance improvement on recommender systems in terms of prediction accuracy applied to e-commerce datasets

Users need to find easily the items that meet their tastes from the massive amount of items. However, for example, on e-commerce websites, there are many products. To find the relevant one for them is not an easy task. Using recommendation algorithms efficiently, information retrieval can be done in the most accurate way. In this work, we try to find out how to deliver the most relevant items to users' interests using different machine learning algorithms and methods.

Speaker: ???

Date: 17/10/2018

Research Seminar 1

2018-2019: Research Seminar 1

Predicting what happens to critically ill patients in hospital with machine learning.

Hospitals use various different scoring systems for patient monitoring that take patient vital signs as their input- usually to decide level of care needed. I explore how using one of the most recent scoring systems can be used in various machine learning techniques to predict whether a patient lives or dies (and more!). To find out which machine learning technique was used-and the results- attend!

Speaker: ???

Date: 10/10/2018