Adapting a classification rule to local and global shift when only unlabelled data are available

Hofer, Vera

doi:10.1016/j.ejor.2014.11.022

Cited by 15 publications

(21 citation statements)

References 37 publications

(71 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Topical research question include, for example, how to update PD scorecards in the face of new information (Hofer, 2015;Sohn & Ju, 2014). The prevailing methods to develop PD models are classification and survival analysis.…”

Section: Literature Reviewmentioning

confidence: 99%

Benchmarking state-of-the-art classification algorithms for credit scoring: An update of research

Lessmann

Baesens

Seow

et al. 2015

European Journal of Operational Research

848

568

View full text Add to dashboard Cite

Section: Literature Reviewmentioning

confidence: 99%

Benchmarking state-of-the-art classification algorithms for credit scoring: An update of research

Lessmann

Baesens

Seow

et al. 2015

European Journal of Operational Research

848

568

View full text Add to dashboard Cite

“…Whereas many research papers address the problem of adapting a classification or regression model after a drift (Hofer, 2015;Hofer & Krempl, 2013 and the literature therein), change point detection focuses on finding time points at which a sequence of random variables or time series undergoes significant changes in characteristics. This question typically arises in quality control where an alarm ought to be set off as early as possible when the process is 'out-of-control' (Qiu, 2013, p. 225).…”

Section: Related Workmentioning

confidence: 99%

Statistical monitoring for continual quality control of railway ballast

Hofer

Bach²

2015

Expert Systems with Applications

Self Cite

View full text Add to dashboard Cite

“…In 2D images where pixels are features, ground distance has a natural physical meaning. In Hofer [62], Euclidean distance in the feature space is used as the measure for ground distance.…”

Section: Earth Mover Distance (Emd)mentioning

confidence: 99%

Quantification under class-conditional dataset shift

Spence

Inskip

Quadrianto

et al. 2019

Proceedings of the 2019 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining

View full text Add to dashboard Cite

Classification is the estimation of the class of each instance in a dataset, quantification is the estimation of the number of instances of each class in a dataset. Quantification methods typically assume that the data which is being quantified has the same class-conditional distribution as the data on which the quantifier was trained. This thesis addresses the situation where this assumption cannot be made, where there is class-conditional dataset shift between the training data and the test data. The work was motivated by sentiment analysis tasks using tweets on Twitter. By selecting users based on the content of their tweet, the users cannot be considered to have been randomly drawn from the population.In this thesis, domain adaptation methods from classification have been applied to the problem of quantification. Separating the data into explicit sub-domains and quantifying each sub-domain separately can increase quantification accuracy but under certain conditions it can also decrease it. An expression for expected quantification error was derived in closed-form with some simplifying assumptions. In tests on real datasets, a method based on this approach gave a modest improvement to quantification accuracy. Constructing a new feature representation has proved successful for domain adaptation in classification.An approach using Stacked Denoising Autoencoders to generate a new feature representation gave a 3.3% relative improvement in quantification accuracy. Finally, a method based on using Kernel Mean Matching for weighting instances in the training set gave a relative improvement in quantification accuracy of 10.7%. Experiments were conducted on publicly available datasets and also on a custom dataset of Twitter users. I would like to thank... David Weir and Novi Quadrianto for supervising me. Luc Berthouze for chairing my thesis committee. My fellow members of the department, in particular Chris Inskip, Oliver Thomas, Matti Lyra and Miro Batchkarov for helping me get things to work.

show abstract

Adapting a classification rule to local and global shift when only unlabelled data are available

Cited by 15 publications

References 37 publications

Benchmarking state-of-the-art classification algorithms for credit scoring: An update of research

Benchmarking state-of-the-art classification algorithms for credit scoring: An update of research

Statistical monitoring for continual quality control of railway ballast

Quantification under class-conditional dataset shift

Contact Info

Product

Resources

About