Ensemble selection from libraries of models

Caruana, Rich; Niculescu-Mizil, Alexandru; Crew, G.; Ksikes, Alex

doi:10.1145/1015330.1015432

Cited by 625 publications

(500 citation statements)

References 8 publications

Supporting

Mentioning

488

Contrasting

Unclassified

Order By: Relevance

“…With this view in mind, when we are interested in extracting the best features (that is, choose the most informative base classifier values), one way is to do feature selection, where we keep some of the features and discard the rest. Actually, methods where people choose a subset of base classifiers from a large ensemble of candidates [27,34,9,35,11,49,42] do exactly this.…”

Section: Constructing New Uncorrelated Eigenclassifiersmentioning

confidence: 99%

“…Related work 3.5.1. Selecting a subset of classifiers Instead of using all, choosing a subset may lead to higher accuracy and decreased complexity where the idea is to weed out the inaccurate or redundant (having low diversity) classifiers [54,9,35,42]. If the number of base classifiers is not high, a subset can be found by exhaustive search [36].…”

Section: Comparison With Adaboost and Baggingmentioning

confidence: 99%

“…If the number of base classifiers is not high, a subset can be found by exhaustive search [36]. Otherwise heuristic methods are preferred: forward search [9,11,35,49,42], backward search [30,27], genetic algorithms [30,54,35], or optimization algorithms [37,53]. Greedy heuristic algorithms, such as forward and backward search using accuracy as the model selection criteria, are known to find ensembles having accuracies as high as the optimal subset constructed by exhaustive search [35,42].…”

Section: Comparison With Adaboost and Baggingmentioning

confidence: 99%

“…Greedy heuristic algorithms, such as forward and backward search using accuracy as the model selection criteria, are known to find ensembles having accuracies as high as the optimal subset constructed by exhaustive search [35,42]. Forward subset selection has been shown to outperform classical combination schemes such as Bagging and boosting [9].…”

Section: Comparison With Adaboost and Baggingmentioning

confidence: 99%

“…That is, we are interested in both pruning the inaccurate ones and also to keep a check on complexity, we want to prune the redundant. ''Diversity'' measures have been proposed [23,22] and one possibility is to have an incremental, forward search where we add a classifier to an ensemble if it is diverse or adds to accuracy [9,11,35,49,42], or another possibility is to have a decremental, backward search where a classifier is removed or pruned if it is not diverse enough or if its removal does not increase error [30,27]. In this work, we propose an alternative method which combines base classifiers using principal component analysis (PCA) to get uncorrelated eigenclassifiers.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Eigenclassifiers for combining correlated classifiers

Ulas

Yıldız

Alpaydın

2012

Information Sciences

View full text Add to dashboard Cite

Section: Constructing New Uncorrelated Eigenclassifiersmentioning

confidence: 99%

Section: Comparison With Adaboost and Baggingmentioning

confidence: 99%

Section: Comparison With Adaboost and Baggingmentioning

confidence: 99%

Section: Comparison With Adaboost and Baggingmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Eigenclassifiers for combining correlated classifiers

Ulas

Yıldız

Alpaydın

2012

Information Sciences

View full text Add to dashboard Cite

Forecast combination in agricultural economics: Past, present, and the future

Ramsey,

Adjemian

2024

Applied Eco Perspectives Pol

View full text Add to dashboard Cite

Forecasts are common in agricultural settings where they are routinely used for decision‐making. The advent of the computer age has allowed for rapid generation of individual forecasts that can be updated in real time. It is well known that the selection and use of a single forecast can expose the forecaster to serious error as a result of model mis‐specification. Forecast combination avoids this problem by combining information from different forecasts. Although forecast combination can be as simple as averaging across forecasts, advances in machine learning have made it possible to combine forecasts according to more complicated weighting schemes and criteria. We provide an overview of forecast combination techniques, including those at the frontier of current practice and involving machine learning. We also provide a retrospective on the use of forecast combination in agricultural economics and prospects for the future. Several of the techniques are illustrated in an application to forecasting nationwide corn and soybean planted acreage and we demonstrate how forecast combination can improve expert USDA projections.

show abstract

Upper confidence bound integrated genetic algorithm‐optimized long short‐term memory network for click‐through rate prediction

Souissi

Ghorbel

2022

Appl Stoch Models Bus & Ind

View full text Add to dashboard Cite

Data on online advertising is rising rapidly due to the fast development of science and technology. Click-through rate (CTR) prediction has become a critical task regarding the digital advertising industry and a key element in increasing advertising profits and user experience. Therefore, this article describes the problem of CTR prediction as a function of sequence classification tasks. Then, we proposed a novel optimization strategy to solve the high-dimensional problem and find a subset of relevant variables to ensure high performance of our model and maximize the number of clicks. Here, we introduced a feature selection and hyper-parameter optimization approach using genetic algorithms (GA) and the upper confidence bound (UCB) model to optimize micro-targeting technology, along with the long short-term memory (LSTM) network-based CTR prediction model. The efficiency of the proposed UCB-LSTM-GA model and two hybrid models, namely LSTM-GA and LSTM-PSO, is evaluated by comparing them to each other and to other machine-learning-based classification methods, including LSTM using a UCB algorithm (UCB-LSTM), High-order Attentive Factorization Machine (HoAFM), genetic algorithm-artificial neural network (GA-ANN), and a feature interaction graph neural network model (Fi-GNN). Our solution achieved as high as 87%, 89%, and 92% for respectively accuracy, precision, and recall, using the popular python tools with real Avazu datasets.

show abstract

Ensemble selection from libraries of models

Cited by 625 publications

References 8 publications

Eigenclassifiers for combining correlated classifiers

Eigenclassifiers for combining correlated classifiers

Forecast combination in agricultural economics: Past, present, and the future

Upper confidence bound integrated genetic algorithm‐optimized long short‐term memory network for click‐through rate prediction

Contact Info

Product

Resources

About