Dr Gavin Smith is an associate professor at The University of Nottingham, where he is the data science lead within the N/LAB. Dr. Smith holds a PhD in Computer Science and specialises in machine learning and (time series) data mining with a focus on data driven algorithms and metrics that facilitate decision making in real world environments. He has published internationally in a range of multidisciplinary venues including Neural Information Processing Systems (NeurIPS), Nature Communications, Scientific Reports, IEEE International Conference on Data Mining (ICDM), IEEE International Conference on Pervasive Computing and Communications (PERCOM), the IEEE International Conference on Big Data and has presented as an invited speaker at events such as the 2017 Korea Green Innovation Days focused on sustainable development in Dar es Salaam, Tanzania.
Gavin's research focuses on understanding human behaviour, where he primarily specializes in two application areas: (1) data analytics/mining/machine learning for international development and (2) the more traditional field of consumer analytics, both of which he has worked in for over 13 years with projects involving over 20 organizations and four continents.
Examples of recent projects related to international development include the generation of origin-destination matrices from anonymized telecommunication data for use in generating citywide transport master-plans (with the World Bank), automated detection of road quality from satellite imagery to free up resources within the Department of Road Zanzibar (with FTL & DFID) and the development of maps of economic potential based on the analysis of mobile money in order to better locate stores in order to aid retail management in planning in developing countries (with the Bill & Melinda Gates Foundation).
From a more academic point of view his research focuses on explainable AI in general (i.e. using novel techniques in conjunction with predictive models to understand phenomena of interest, rather than just an arbitrary individual model!) as well as the efficient and effective processing of large time series data sets in order to better understand/summarise/predict human behaviour, addressing issues such as non-stationarity, intermittence (event series vs traditional time series) while ensuring appropriate interpretability for real world use. He has previously worked on projects on pedestrian route prediction, image segmentation and the validity of implicitly crowd-sourced big data.
D. BARRACK, J. GOULDING, K. HOPCRAFT, S. PRESTON and G. SMITH, 2015. AMP: a new time-frequency feature extraction method for intermittent time-series data. In: SIGKDD Workshop on Mining and Learning from Time Series. [arXiv.org]
J. ROSSER, J.MORLEY and G. SMITH, 2015. Modelling of Building Interiors with Mobile Phone Sensor Data. ISPRS Int. J. Geo-Inf. 2015, 4, 989-1012. [Via Journal, open access]
J. PINCHIN, G. SMITH, C. HILL, T. MOORE and I. LORAM, 2014. The Potential of Electromyography to Aid Personal Navigation In: Proceedings of the 27th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS 2014), Tampa, Florida. [pre-print ION Publications]
G. SMITH, R. WIESER, J. GOULDING and D. BARRACK, 2014. A Refined Limit on the Predictability of Human Mobility In: IEEE International Conference on Pervasive Computing and Communications (PerCom), Budapest. [supporting site pre-print IEEE Xplore]
G. SMITH, J. GOULDING and D. BARRACK, 2013. Towards optimal symbolization for time series comparisons In: IEEE International Conference on Data Mining Workshops (ICDMW), Dallas. [pre-print IEEE Xplore]
M. DIMOND, G. SMITH and J. GOULDING, 2013. Improving Route Prediction through User Journey Detection In: ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems, Orlando. [ACM Digital Library]
G. SMITH, C. BRIAN and H. ASHMAN, 2012. Evaluating implicit judgements from image search clickthrough data In: Journal of the American Society for Information Science and Technology 63, 12 (December 2012). [Avaliable via Journal]
G. SMITH and J MORLEY, 2012. Full floor identification in images with minimal close range 3D information In: IEEE International Conference on Image Processing. [Demo code IEEE Xplore]
J. GOULDING, G. SMITH and D. BARRACK, 2012. Neo-demographics and Distributions in the Digital Shadow In: Third Annual Digital Economy All Hands Conference.
M. DIMOND, G. SMITH, J. GOULDING, M. JACKSON and X. MENG, 2012. Comparing predefined and learned trajectory partitioning with applications to pedestrian route prediction In: GISRUK 2012, 20th annual GIS Research UK.
G. SMITH and J. GOULDING, 2011. Pedestrian Route Prediction from GPS Logs using Augmented Cover Trees In: ECML-PKDD: International Workshop on Finding Patterns of Human Behaviors in Networks and Mobility Data.
H. ASHMAN, M. ANTUNOVIC, S. CHAPRASIT, G. SMITH and M. TRURAN, 2011. Implicit association via crowd-sourced coselection. In: Proceedings of the 22nd ACM conference on Hypertext and hypermedia ACM.
G. SMITH and H. ASHMAN, 2009. Evaluating implicit judgements from image search interactions In: Web Science Conference.
H. ASHMAN, M. ANTUNOVIC, C. DONNER, R. FRITH, E. REBELOS, J-F. SCHMAKEIT, G. SMITH and M. TRURAN 2009. Are clickthroughs useful for image labelling? In: Proceedings of the International Conference on Web Intelligence, Milan.
G. SMITH, T. BRAILSFORD, C. DONNER, D. HOOIJMAIJERS, M. TRURAN, J. GOULDING and H. ASHMAN, 2009. Generating unambiguous URL clusters from Web search In: Proceedings of the Workshop on Web Search Click Data, Barcelona.
G. SMITH and G. WIGLEY, 2008. High Level Languages for Reconfigurable Computing: An Equivalent to Third Generation Software Lan- guages? In: Conference on Engineering of Reconfigurable Systems and Algorithms, Las Vegas.