News by the Artificial Intelligence Group
Research seminar "Machine Learning in the Travel Industry" by Pedro Ferreira
Machine Learning has recently been applied in a Knowledge
Transfer Partnership between UWE Bristol and Paxport/Multicom, a
leading provider of online booking of travel and holidays in the UK
Machine Learning techniques based on Collaborative Filtering have been applied to increase the perceived relevance of user search results, particularly when searching for accommodation as part of holiday bookings.
It was found that, although historical search data is abundant, it is not possible to identify end users, and poor data quality and sparsity also present significant challenges. To address these challenges, collaborative filtering techniques such as implicit-Alternating Least Squares (iALS) have been used to identify patterns in the data, and latent Dirichlet allocation (LDA) has been exploited to derive semantic topics from textual descriptions of accommodation to create structured information in the form of topic distribution vectors, opening up possibilities for new techniques and allowing for improved analysis over the data.
Initial results show a promising increase in search relevance, shedding new light on the trends in the historical search data and providing knowledge discovery of accommodation bookings for Paxport/Multicom.
Watch the recorded seminar by Pedro Ferreira, held on 5 April 2017, at UWE Bristol.
Text caption support is provided for this video on YouTube.
Research Associate position
As a Research Associate, you will join UWE
Bristol's Artificial Intelligence Group to explore novel
combinations of machine learning, meta-heuristic optimisation and
graph partitioning, for tackling large scale problems, such as
those arising from our collaborations with external partners such
as the Office for National Statistics.
With a Masters Degree in Artificial Intelligence (or similar), you will have an understanding of metaheuristics and machine leaning, and possess the skills needed to produce high quality open source software suitable for use by collaborators.
In this role, you will work independently, and with members of the Artificial Intelligence Group and its external partners, to conduct research into how ideas from hyper-heuristics, surrogate modelling and graph partitioning can be applied to improve metaheuristic approaches to large scale problems, and undertake activities to maximise the impact of research findings.
The application deadline for this position has now passed.
Best Paper Award at the CyberSA 2016 conference
The work of Dr Phil Legg, of the AI group at UWE Bristol, was recently awarded Best Paper at the International Conference on Cyber Situational Awareness, Data Analytics and Assessment (CyberSA) 2016.
His paper, “Enhancing Cyber Situation Awareness for Non-Expert Users using Visual Analytics” addressed the need for developing intuitive visual interfaces for being able to quickly and easily identify network traffic and activities.
In particular, whilst many security experts may be familiar with deep exploration, the paper discusses the ways non-experts can be encouraged to start exploring how their devices, such as smartphones and laptops, communicate over wireless Internet connections.
The paper positions this new area of research within the wider context of security analytics, and presents a visual analytics dashboard designed to be intuitive for non-experts, along with a use case of identifying malicious network traffic from a malware infection. Future work will study how non-experts respond to such interfaces, and this award demonstrates that the research community value this as a promising area of research to be developed.
New studentship available
The AI group has an opportunity for a
fully-funded PhD studentship, working at the cutting edge of
Artificial Intelligence and 'Big-Data', to be co-funded by UWE
Bristol and Montvieux Ltd.
The application deadline for this studentship has now passed.
New book published
Springer Verlag have published the new edition of Jim Smith and Gusz Eiben’s best selling textbook “Introduction to Evolutionary Computing”.
This new edition of their well-established undergraduate textbook has been revised to offer an integrated view on evolution-based problem-solving algorithms.
The AI group had a research position available on big data analytics and query relevance optimization, working as an associate on a Knowledge Transfer Partnership with Multicomm Ltd. The application deadline for this position has now passed.
Paper accepted in Nature!
Jim Smith and Gusz Eiben’s paper “From Evolutionary Computation to the evolution of things” has been accepted to appear in the prestigious journal Nature later in May.
The AI group has been awarded funding of £107K for KTP project with MultiComm ltd using metaheuristics and machine learning to improve the accuracy and relevance of the results returned by their engine which powers many online travel sites in the UK.
Two new members of staff
Welcome to two new members of staff to the group: Phil Legg (Visualisation techniques) and Mehmet Aydin (Industrial Optimisation), and a visiting PhD Student from University of Sao Paolo (Daniel Pigatto).
Chris Simons presents on Interactive Evolutionary Algorithms at the ACCU Conference for 2015
Chris Simons presented a workshop on Interactive Evolutionary Algorithms to programmers and software developers at the 2015 annual conference of the Association of C and C++ Users (ACCU), held between 22 and 27 April, 2015. The workshop was entitled “Evolving Software: A Fusion of the Developer and Evolutionary Algorithms” and explored novel ways in which AI, specifically Evolutionary Algorithms, can be exploited in software developer practice and toolsets.
Dr Phil Legg gives invited talk on Insider Threat Mitigation at the University of Oxford
Dr. Phil Legg gave an Invited Talk at the 2nd UK Workshop on Insider Threat Mitigation, which was hosted by the University of Oxford on Friday, 20 March 2015. The talk, entitled, “Caught in the Act of an Insider Attack: Detection and Assessment of Insider Threat”, investigated how machine learning and data visualisation can be utilised to identify anomalous users in the context of insider threat detection. The full paper was published at the 2015 IEEE International Symposium on Technologies for Homeland Security.
Call for Papers for the 2015 Metaheuristic Design Patterns Workshop
The call for papers at the 2015 Metaheuristic Design Patterns workshop at GECCO 2015 has gone out.
Post available as graduate tutor with research into AI/Forensic computing
The department advertised vacancies for graduate tutors - mixing teaching with post graduate research study. One of the posts will be in the area of applying AI to problems in internet security.
The closing date for applications has now passed.
Jim Smith appointed as Professor of Interactive Artificial Intelligence
In recognition of the importance of this growing area UWE has created a new post of Professor of Interactive Artificial Intelligence. After external advertising, Jim Smith was appointed from 1st April.
Chris Simons examines PhD students in Sweden
Chris Simons acted as external assessor for PhD candidate in Blekinge Institute of Technology, Sweden, building valuable links with the group there, working on interactive search-based software engineering.
Jim Smith invited to be keynote speaker
Jim Smith was invited to be keynote speaker at IEEE Symposium Series on Computational Intelligence, Florida (2014).
Building links with China
Jim Smith is International Advisor to the Memetic Computing Lab (McLab) at the School of Computer Science, China University of Geosciences.