Home Page of Bob Durrant

About me:

I am a lecturer in the Department of Statistics at the University of Waikato, New Zealand. I joined the department on 1st July 2013.

I completed my PhD at the School of Computer Science, University of Birmingham, U.K. where I was supervised by Ata Kaban. My PhD research focused on mitigating issues associated with working with very high-dimensional data by using random projections, and quantifying their impact on classification performance. I submitted my thesis on the 23rd January 2013 and my viva took place on 29th April 2013, when my examiners were John Shawe-Taylor and Peter Tino.

I spent a 4 month stint as a Research Fellow in the School of Computer Science at Birmingham before taking up my current post. Prior to that, in 2008-9 I was a Teaching Fellow in the school and taught masters and undergraduate modules on Machine Learning and Nature Inspired Design. I was also admissions tutor for MSc Natural Computation and MSc Intelligent Systems Engineering.

Before joining academia I had a varied career in industry, most recently as a project manager for a large UK stockbroker.

I am a member of the New Zealand Statistical Association, the Australia and New Zealand chapter of SIGKDD, and a professional member of the Royal Society of New Zealand.

My Erdös number is 5.



Contact Details:

Robert J. Durrant, Room G.3.30,

Department of Statistics,

University of Waikato,

Private Bag 3105,

Hamilton 3240

New Zealand

e: bobd [at] waikato [dot] ac [dot] nz

w: http://www.stats.waikato.ac.nz/~bobd/

t: +64 (0)7 838 4466 x8334

f: +64 (0)7 838 4155



Research Interests

I have a broad interest in learning and vision, both the organic and machine varieties.

I have a particular interest in computational theories of learning.

My recent research has been into dimensionality reduction. In particular, random projections of very high dimensional data sets into low dimensional spaces, and the effect of projection on classification performance.

Currently I am working on better understanding the effect of randomization in the context of classifier ensembles. In particular whether, and how, randomization improves generalization when the number of training examples is much smaller than the dimensionality of the data.

I am also interested in the Measure Concentration phenomenon and its application to machine learning and statistical pattern recognition theory and practice. Along similar lines I have some (still largely unexplored) interest in randomized algorithms.



Publications and Presentations

Journal Papers:

R.J. Durrant and Ata Kaban. Random Projections as Regularizers: Learning a Linear Discriminant from Fewer Observations than Dimensions. (Machine Learning - To Appear). pdf

A. Kaban, J. Bootkrajang, and R.J. Durrant. Towards Large Scale Continuous EDA: A Random Matrix Theory Perspective. (Under Revision). pdf code

R.J. Durrant and Ata Kaban. A tight bound on the performance of Fisher's linear discriminant in randomly projected data spaces. Pattern Recognition Letters, 33(7), pp 911 – 919, 2012. pdf

R.J. Durrant and Ata Kaban. When Is 'Nearest Neighbor' Meaningful: A Converse Theorem and Implications. Journal of Complexity, 25(4), pp 385-397, August 2009. pdf



Refereed Conference Papers:

R.J. Durrant and A. Kaban. Random Projections as Regularizers: Learning a Linear Discriminant Ensemble from Fewer Observations than Dimensions. Proceedings 5th Asian Conference on Machine Learning (ACML 2013). JMLR W&CP 29 : pp 17-32, 2013.(Best paper award). pdf supp

A. Kaban and R.J. Durrant. Dimension-adaptive bounds on Compressive FLD Classification. Proceedings 24th International Conference on Algorithmic Learning Theory (ALT 2013 ). LNAI 8139, pp 294-308. Springer. pdf

R.J. Durrant and A. Kaban. Sharp Generalization Error Bounds for Randomly-projected Classifiers. Proceedings 30th International Conference on Machine Learning (ICML 2013). JMLR W&CP 28(3): pp 693-701, 2013. pdf

A. Kaban, J. Bootkrajang, and R.J. Durrant. Towards Large Scale Continuous EDA: A Random Matrix Theory Perspective. Proceedings Genetic and Evolutionary Computation Conference (GECCO 2013), pp 383-390. ACM. (Best paper award in the GDS/EDA track). pdf code

R.J. Durrant and A. Kaban. Error bounds for Kernel Fisher Linear Discriminant in Gaussian Hilbert Space. Proceedings 15th International Conference on Artificial Intelligence and Statistics (AIStats 2012). JMLR W&CP 22: pp 337-345, 2012. pdf

R.J. Durrant and A. Kaban. Compressed Fisher Linear Discriminant Analysis: Classification of Randomly Projected Data. Proceedings 16th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD 2010), pp 1119-1128. ACM. pdf

R.J. Durrant and A. Kaban. A bound on the performance of LDA in randomly projected data spaces. Proceedings 20th International Conference on Pattern Recognition (ICPR 2010), pp 4044-4047. IEEE. (IBM Best Student Paper Award in the Pattern Recognition and Machine Learning track). pdf

A. Kaban and R.J. Durrant. Learning with Lq<1 vs L1-norm regularization with exponentially many irrelevant features. Proceedings 19th European Conference on Machine Learning (ECML08). LNAI 5211, pp. 580-596. Springer. pdf

A. Kaban and R.J. Durrant. A norm-concentration argument for non-convex regularization. ICML/UAI/COLT Workshop on Sparse Optimization and Variable Selection, 9 July, 2008, Helsinki, Finland. Slides (pdf)



PhD Thesis:

R.J. Durrant. Learning in High Dimensions with Projected Linear Discriminants. pdf



Technical Reports:

R.J. Durrant and Ata Kaban. Flip Probabilities for Random Projections of θ-separated vectors. University of Birmingham, School of Computer Science Technical Report CSR-10-10. pdf

R.J. Durrant and Ata Kaban. A comparison of the moments of a quadratic form involving orthonormalised and normalised random projection matrices. University of Birmingham, School of Computer Science Technical Report CSR-11-04. pdf



Invited Tutorial:

Random Projections for Machine Learning and Data Mining: Theory and Applications. ECML-PKDD 2012, Bristol. Slides (pdf) Handouts (6 Slides per page) (pdf)



Invited Poster Presentation:

Sparsity in the context of learning from high dimensional data - with Ata Kaban. ICARN International Workshop 26 Sept 2008, Liverpool. pdf



Conference Poster:

Finite Sample Effects in Compressed Fisher's LDA - with Ata Kaban. 'Breaking News' poster presented at the 13th International Conference on Artificial Intelligence and Statistics (AIStats 2010) Poster (pdf) Proof (pdf)



Talks:

Random Projections as Regularizers: Learning a Linear Discriminant from Fewer Observations than Dimensions. Statistics Seminar, University of Auckland, 6th August 2014;

What's your angle? Are most triangles acute or obtuse? Community Open Day Talk, University of Waikato, 17 May 2014.

Random Projections for Data Mining and Optimization: Theory and Applications. NZSA ORSNZ Joint Conference, University of Waikato, 27th November 2013;

Why thousand-dimensional vector spaces are interesting. Mathematics Teachers' Conference, School of Mathematics, University of Birmingham, 7 July 2011.

Random triangles and the curse of dimensionality. Postgraduate seminar, School of Computer Science, University of Birmingham, 2 March 2011; Workshop presentation at 6th-form Mathematics Conference, King Edward's Camp Hill School for Girls, Birmingham, 4 March 2011.

Compressed Fisher Linear Discriminant Analysis: Classification of Randomly Projected Data. Departmental Seminar, School of Computer Science, University of Birmingham, 10 June 2010.



Reviewing:

Journals:

Transactions on Pattern Analysis and Machine Intelligence (IEEE)

Transactions on Neural Networks and Learning Systems (IEEE)

Transactions on Systems, Man and Cybernetics (IEEE)

IEEE Computational Intelligence Magazine - Special Issue on Computational Intelligence in Big Data

Pattern Analysis and Applications (Springer)

Journal of Classification (Springer)

Pattern Recognition Letters (Elsevier)

Neurocomputing (Elsevier)

Information Sciences (Elsevier)

Conferences and Workshops:

1st International Workshop on High-dimensional Data Mining at IEEE ICDM 2013

Special session on Label Noise in Classification at ESANN 2014



Program Committees:

1st International Workshop on High-dimensional Data Mining at IEEE ICDM 2013

6th Asian Conference on Machine Learning (ACML 2014)



Last Revised 18/06/14