Data Mining and Classification

Classification

is the task of assigning records in a data set to one of a finite set of classes.

Prediction (also know as estimation or regression)

is the assumption that the future trend of variations in the value of a time-series variable will mirror the trend of variations in the value of the same variable for similar historical time series.

We have used the AURA modular, binary neural architecture to implement a k-Nearest Neighbour (k-NN) classifier [HA_KAIS05, HA_TR2012, HJA_FIATS12, KHetal_ITS10, KHAPL_TRB10, KHAP_UTSG10, HKAP_TT10, HLA_NN04, WHOAL_IWANN03]. k-NN is a method for classifying objects based on the closest matching examples (neighbours) in the feature space. The training examples are taken from a set of objects for which the correct classification (or, in the case of prediction, the value of the property) is known. k-NN is a type of instance-based learning where the function is only approximated locally and all computation is deferred until classification or prediction.

The AURA k-NN classifier has been adapted for short-term prediction to proactively manage traffic to allow us to anticipate traffic problems before they occur. k-NN simply sets the value of the variable for the object (the prediction at t+n) to be the average of the values of its k nearest neighbours at t+n. The AURA k-NN predictor [HJA_NCTA12, HJA_FIATS12, HKetal_UTSG11] incorporates both spatial and temporal characteristics into the prediction process. The algorithm combines information from multiple traffic sensors (spatial lag) and time-series data (temporal lag).

Our research also covers the use of attribute selection and attribute weighting in classifiers (or predictors) [HJA_NCTA12, HOA_BICS04, HOA_NC06] to allow the salient attributes to be pinpointed and weighted according to their respective importance to the classification goal. This is particulalry germane in high dimensional data sets where many attributes may be redundant and their removal will both increase classification or prediction accuracy and speedup the recall process.

We have extended the AURA k-NN technique to produce a decision table classifier [HOA_BICS04, HOA_NC06]. The AURA k-NN also provides the basis for our shape matching algorithm [HOA_BICS06, HOA_NC09] that represents shapes as binary codes and then matches the shape representations within the AURA k-NN framework.

We have also extended the AURA k-NN classifier technique to allow outlier detection [OutlierBook, HA_SAGE13HA_AIRE05] by identifying records in a particular data set which are distant from the majority of the population [ABJH_CM10]. These outlying records may indicate novelty, anomalous readings, extreme-valued records, system faults or even critical failures so they need to be detected to allow them to be dealt with appropriately.

Books

Victoria J. Hodge, [Hodge_OutlierBook].
Outlier and Anomaly Detection:
A Survey of Outlier and Anomaly Detection Methods,
Lambert Academic Publishing: ISBN 978-3-8465-4822-6, 2011

Victoria J. Hodge, [Hodge_BOOK].
Integrating Information Retrieval with Artificial Neural Networks:
Implementing a Modular Information Retrieval System using Artificial Neural Networks,
Lambert Academic Publishing: ISBN 978-3-8433-7966-3, 2010

Journal Articles, Book Chapters, Conference Proceedings, Technical Reports, Thesis.

Unfortunately copyright restrictions prevent me from making some of my publications available on-line. However, reprints are available from me on request.  More Info...

2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

Victoria J. Hodge & Jim Austin [HA_SAGE13].
A Survey of Outlier Detection Methodologies.
In S. Babones (Ed.), Fundamentals of Regression Modeling, SAGE Publications, 2013. (Original work published 2004).
Publisher's website.

Victoria J. Hodge, Tom Jackson & Jim Austin [HJA_NCTA12].
A Binary Neural Network Framework for Attribute Selection and Prediction.
In, Proceedings of the 4th International Conference on Neural Computation Theory and Applications (NCTA 2012), (pp. 510-515), Barcelona, Spain, October 5-7, 2012: SciTePress.
download pdf (.pdf).

Richard Mounce, Garry Hollier, Mike Smith, Victoria J. Hodge, Tom Jackson & Jim Austin [TRC_2012].
A metric for pattern-matching applications to traffic management.
Transportation Research Part C: Emerging Technologies, Elsevier, Available online 12 June 2012.
download final submitted pdf (.pdf).

Victoria J. Hodge & Jim Austin [HA_TR2012].
Discretisation of Data in a Binary Neural k-Nearest Neighbour Algorithm.
Technical Report YCS-2012-473, Department of Computer Science, University of York, UK, June 2012.
download pdf (.pdf).

Victoria J. Hodge, Tom Jackson & Jim Austin [HJA_FIATS].
Intelligent Decision Support using Pattern Matching.
In, Proceedings of the 1st International Workshop on Future Internet Applications for Traffic Surveillance and Management (FIATS-M 2011), Sofia, Bulgaria, Oct 2011. ISBN:978-989-8425-87-4.
download pdf (.pdf).

Victoria J. Hodge, Rajesh Krishnan, Tom Jackson, Jim Austin & John Polak [HKetal_UTSG11].
Short-Term Traffic Prediction Using a Binary Neural Network.
Presented at, 43rd Annual UTSG Conference, Open University, Milton Keynes, UK, January 5-7, 2011.
download pdf (.pdf).

Rajesh Krishnan, Victoria J. Hodge, Jim Austin, John Polak, Tom Jackson, Mike Smith & Tzu-Chang Lee [KHetal_ITS10].
Intelligent Decision Support for Traffic Management.
In, Proceedings of 17th ITS World Congress: (CD-ROM), Busan: Korea, October 25-29, 2010.
download pdf (.pdf).

Jim Austin, Grant Brewer, Tom Jackson & Victoria J. Hodge [ABJH_CM10].
AURA-Alert: The use of binary associative memories for condition monitoring applications.
In, Proceedings of 7th International Conference on Condition Monitoring and Machinery Failure Prevention Technologies: (CM 2010 and MFPT 2010), Stratford-upon-Avon, England, 22-24 June, 2010. ISBN: 978-1-61839-013-4, Vol. 1: pp. 699-711.
download pdf (.pdf).

Victoria J. Hodge, Rajesh Krishnan, Jim Austin & John Polak [HKAP_TT10].
A computationally efficient method for online identification of traffic incidents and network equipment failures.
Presented at, Transport Science and Technology Congress: TRANSTEC 2010, Delhi, April 4-7, 2010.
download pdf (.pdf).

Rajesh Krishnan, Victoria J. Hodge, Jim Austin, John Polak & Tzu-Chang Lee [KHAPL_TRB10].
On Identifying Spatial Traffic Patterns using Advanced Pattern Matching Techniques.
In, Proceedings of Transportation Research Board (TRB) 89th Annual Meeting, Washington, D.C., January 10-14, 2010.
(DVD-ROM: 2010 TRB 89th Annual Meeting: Compendium of Papers).

Rajesh Krishnan, Victoria J. Hodge, Jim Austin & John Polak [KHAP_UTSG10].
A Computationally Efficient Method for Online Identification of Traffic Control Intervention Measures.
Presented at, 42nd Annual UTSG Conference, Centre for Sustainable Transport, University of Plymouth, UK: January 5-7, 2010.
download pdf (.pdf).

Victoria J. Hodge, Simon O'Keefe & Jim Austin [HOA_NC09].
A Binary Neural Shape Matcher using Johnson Counters and Chain Codes.
NeuroComputing, 72(2009): pp. 693-703, Elsevier Science, 2009.
Full Text Article from Elsevier Science Journals – NeuroComputing (pdf).

Victoria J. Hodge, Simon O'Keefe & Jim Austin [HOA_BICS06].
A Binary Neural Shape Matcher using Johnson Counters and Chain Codes.
In, Brain Inspired Cognitive Systems 2006 (BICS 2006), Island of Lesvos, Greece. October 10-14, 2006.
download pdf (.pdf).

Victoria J. Hodge, Simon O'Keefe & Jim Austin [HOA_NC06].
A Binary Neural Decision Table Classifier.
NeuroComputing, 69(16-18): pp. 1850-1859, October 2006.
(Selected papers from the 1st International Conference on Brain Inspired Cognitive Systems).
Full Text Article from Elsevier Science Journals – NeuroComputing (pdf).

Victoria J. Hodge & Jim Austin [HA_KAIS05].
A Binary Neural k-Nearest Neighbour Technique.
Knowledge and Information Systems, 8(3): pp. 276-292, Springer-Verlag London Ltd, 2005.
download final submitted version (.pdf).

Victoria J. Hodge & Jim Austin [HA_AIRE04].
A Survey of Outlier Detection Methodologies.
Artificial Intelligence Review, 22: pp. 85-126, Kluwer Academic Publishers, 2004.
download final submitted version (.pdf).

Victoria J. Hodge, Simon O'Keefe & Jim Austin [HOA_BICS04].
A Binary Neural Decision Table Classifier.
In, Proceedings Brain Inspired Cognitive Systems 2004 (BICS 2004), University of Stirling, Scotland, UK. August 29 to September 1, 2004. ISBN: 1-85769-199-7.
download pdf (.pdf).

Victoria J. Hodge, Ken Lees & Jim Austin [HLA_NN04].
A High Performance k-NN Approach Using Binary Neural Networks.
Neural Networks, 17(3): pp. 441-458, Elsevier Science, 2004.
download final submitted version (pdf (.pdf)).

Michael Weeks, Vicky Hodge, Simon O'Keefe, Jim Austin & Ken Lees [WHOAL_IWANN03].
Improved AURA k-Nearest Neighbour Approach.
In, Proceedings of IWANN-2003, International Work-conference on Artificial and Natural Neural Networks, Mahon, Menorca, Balearic Islands, Spain, June 3-6, 2003.
Lecture Notes in Computer Science (LNCS) 2687, Springer Verlag, Berlin.
download pdf (.pdf).