Theoretical and empirical analysis of diversity in non-stationary learning

Stapenhurst, Richard; Brown, Gavin

doi:10.1109/cidue.2011.5948488

Cited by 4 publications

(3 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Conversely, weightless frameworks emphasize plasticity and tend to drop weaker ensemble members immediately (e.g., [81,167]); • Classifier weight adaption versus data instance based weight adaptation: The weighing of votes from an ensemble is generally a function of either classifier performance [54,65,152] or of the data from which a member of the ensemble was constructed (e.g., [20,150]); • Identification of ensemble member for replacement: Various heuristics have been proposed for targeting the ensemble member for replacement when performance as a whole is deemed to be poor, e.g., replace the oldest [167] or member with least 'contribution' [107,171]. • Role of diversity on ensembles: Within an environment undergoing change, diversity provides faster reaction times to a change, but does not necessarily facilitate fast convergence to the new concept [137,165]. One implication of this might be that the amount of diversity/plasticity needs to in some way 'match' the amount of concept drift/shift in the stream.…”

Section: Ensemble ML Perspectivementioning

confidence: 99%

“…The diversity term can be characterized as being 'good' or 'bad' [26], a result that has a corresponding observation in EC [187]. Constructing new ensemble member Diversity of base learner [65,102] Sample stream data using Boosting versus Bagging [147,150] Identify ensemble member for replacement Age based heuristics [167] Performance based heuristics [107,171] Class imbalance Effect of sampling biases [34,55,86,186] Drift management Incremental updating of current models [107,158] Adapt voting weights [2,90,152] Shift management Outright replacement of one or more ensemble member [68,81,167] Diversity management Impact on capacity for change [26,137,165] Genet Program Evolvable Mach Under non-stationary data, it has been established that reducing the absolute value for the ensemble margin produces an equivalent increase in diversity [165]. A second open question is in regard to the method assumed for combining the outcome from multiple models under a non-stationary task.…”

Section: Ensemble ML Perspectivementioning

confidence: 99%

See 1 more Smart Citation

Evolutionary model building under streaming data for classification tasks: opportunities and challenges

Heywood

2014

Genet Program Evolvable Mach

View full text Add to dashboard Cite

Streaming data analysis potentially represents a significant shift in emphasis from schemes historically pursued for offline (batch) approaches to the classification task. In particular, a streaming data application implies that: (1) the data itself has no formal 'start' or 'end'; (2) the properties of the process generating the data are non-stationary, thus models that function correctly for some part(s) of a stream may be ineffective elsewhere; (3) constraints on the time to produce a response, potentially implying an anytime operational requirement; and (4) given the prohibitive cost of employing an oracle to label a stream, a finite labelling budget is necessary. The scope of this article is to provide a survey of developments for model building under streaming environments from the perspective of both evolutionary and non-evolutionary frameworks. In doing so, we bring attention to the challenges and opportunities that developing solutions to streaming data classification tasks are likely to face using evolutionary approaches.

show abstract

Section: Ensemble ML Perspectivementioning

confidence: 99%

Section: Ensemble ML Perspectivementioning

confidence: 99%

Evolutionary model building under streaming data for classification tasks: opportunities and challenges

Heywood

2014

Genet Program Evolvable Mach

View full text Add to dashboard Cite

show abstract

“…Diversity and voting margins are two important aspects of ensemble methods that are theoretically connected. The voting margin can be defined as the confidenceweighted correctness of the prediction on a given example (Richard Stapenhurst, 2011). Voting margins are useful in determining generalization error.…”

Section: Background On Ensemble Diversity and Voting Marginsmentioning

confidence: 99%

The impact of ensemble diversity on learning big data in dynamic environments

Museba

Kalpa Publications in Computing

View full text Add to dashboard Cite

For many classification tasks, data is collected over an extended period of time and the predictive model learns over time, adapting to changes in the underlying distribution of the data if necessary. To optimize generalization performance, margin distribution is considered to be an important factor. A major concern posed by nonstationary learning for any algorithm is the rate of adaptation to new concepts and the volume of the data. Tackling the problem of learning in nonstationary environments associated with drifting concepts with ensembles of classifiers makes the concept of diversity to be of paramount significance in optimizing the rate of adaptation to new concepts for classification tasks. In this paper, we investigate the impact of ensemble diversity on the rate of adaptation to new concepts in nonstationary learning. The rate of adaptation is analyzed by exploiting the correspondence that exists between voting margins and the double fault measure, a popular diversity measure strongly linked to the margin. We utilize the Adaptive Classifier Ensemble Boost algorithm (AceBoost) to generate diverse base classifiers and optimize margin distribution to exploit different amounts of diversity to generate an optimal ensemble capable of handling different kinds of drift. The experimental results confirm that AceBoost outperforms other state of the art algorithms that exploit ensemble diversity to handle concept drift.

show abstract