Sequential Feature Selection for Classification

Rückstieß, Thomas; Osendorfer, Christian; Smagt, Patrick van der

doi:10.1007/978-3-642-25832-9_14

Cited by 62 publications

(45 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A sequential backward selection [11] strategy was applied to carefully select a small group of significant features from X . Then, an inner SVM was wrapped into the feature selection framework to evaluate the predictive accuracy for candidate subset of features using a leave-one-out cross validation.…”

Section: Methodsmentioning

confidence: 99%

Outcome Prediction for Patient with High-Grade Gliomas from Brain Functional and Structural Networks

Liu

Zhang

Rekik

et al. 2016

Lecture Notes in Computer Science

View full text Add to dashboard Cite

High-grade glioma (HGG) is a lethal cancer, which is characterized by very poor prognosis. To help optimize treatment strategy, accurate preoperative prediction of HGG patient's outcome (i.e., survival time) is of great clinical value. However, there are huge individual variability of HGG, which produces a large variation in survival time, thus making prognostic prediction more challenging. Previous brain imaging-based outcome prediction studies relied only on the imaging intensity inside or slightly around the tumor, while ignoring any information that is located far away from the lesion (i.e., the “normal appearing” brain tissue). Notably, in addition to altering MR image intensity, we hypothesize that the HGG growth and its mass effect also change both structural (can be modeled by diffusion tensor imaging (DTI)) and functional brain connectivities (estimated by functional magnetic resonance imaging (rs-fMRI)). Therefore, integrating connectomics information in outcome prediction could improve prediction accuracy. To this end, we unprecedentedly devise a machine learning-based HGG prediction framework that can effectively extract valuable features from complex human brain connectome using network analysis tools, followed by a novel multi-stage feature selection strategy to single out good features while reducing feature redundancy. Ultimately, we use support vector machine (SVM) to classify HGG outcome as either bad (survival time ≤ 650 days) or good (survival time >650 days). Our method achieved 75 % prediction accuracy. We also found that functional and structural networks provide complementary information for the outcome prediction, thus leading to increased prediction accuracy compared with the baseline method, which only uses the basic clinical information (63.2 %).

show abstract

Section: Methodsmentioning

confidence: 99%

Outcome Prediction for Patient with High-Grade Gliomas from Brain Functional and Structural Networks

Liu

Zhang

Rekik

et al. 2016

Lecture Notes in Computer Science

View full text Add to dashboard Cite

show abstract

“…The most similar Reinforcement Learning works are the paper by Ji and Carin (2007) and the (still unpublished) paper by Rückstieß et al (2011) which proposes MDP models for cost-sensitive classification. Both of these papers have formalizations that are similar to ours, yet concentrate on cost-sensitive problems.…”

Section: Cost Sensitive Classificationmentioning

confidence: 99%

Sequential approaches for learning datum-wise sparse representations

et al. 2012

View full text Add to dashboard Cite

In supervised classification, data representation is usually considered at the dataset level: one looks for the "best" representation of data assuming it to be the same for all the data in the data space. We propose a different approach where the representations used for classification are tailored to each datum in the data space. One immediate goal is to obtain sparse datum-wise representations: our approach learns to build a representation specific to each datum that contains only a small subset of the features, thus allowing classification to be fast and efficient. This representation is obtained by way of a sequential decision process that sequentially chooses which features to acquire before classifying a particular point; this process is learned through algorithms based on Reinforcement Learning.The proposed method performs well on an ensemble of medium-sized sparse classification problems. It offers an alternative to global sparsity approaches, and is a natural framework for sequential classification problems. The method extends easily to a whole family of sparsity-related problem which would otherwise require developing specific solutions. This is the case in particular for cost-sensitive and limited-budget classification, where feature acquisition is costly and is often performed sequentially. Finally, our approach can handle non-differentiable loss functions or combinatorial optimization encountered in more complex feature selection problems.

show abstract

“…After analyzing the correlations of the variables in the dataset and adding the dichotomic variables for the next analysis, we proceeded to select the variables that influence the best in diagnosing the Metabolic Syndrome to reduce the dimensions of the dataset to a subset. For this purpose, we used Sequential Feature Selection searches for a subset of the features in the full model [47] with comparative predictive power being the variables described in Table 6 in conjunction with the dichotomous variables described in Figure 7.…”

Section: Discussionmentioning

confidence: 99%

Novel Data Mining Methodology for Healthcare Applied to a New Model to Diagnose Metabolic Syndrome without a Blood Test

et al. 2019

View full text Add to dashboard Cite

Metabolic Syndrome (MetS) is a cluster of risk factors that increase the likelihood of heart disease and diabetes mellitus. It is crucial to get diagnosed with time to take preventive measures, especially for patients in locations without proper access to laboratories and medical consultations. This work presented a new methodology to diagnose diseases using data mining that documents all the phases thoroughly for further improvement of the resulting models. We used the methodology to create a new model to diagnose the syndrome without using biochemical variables. We compared similar classification models, using their reported variables and previously obtained data from a study in Colombia. We built a new model and compared it to previous models using the holdout, and random subsampling validation methods to get performance evaluation indicators between the models. Our resulting ANN model used three hidden layers and only Hip Circumference, dichotomous Waist Circumference, and dichotomous blood pressure variables. It gave an Area Under Curve (AUC) of 87.75% by the IDF and 85.12% by HMS MetS diagnosis criteria, higher than previous models. Thanks to our new methodology, diagnosis models can be thoroughly documented for appropriate future comparisons, thus benefiting the diagnosis of the studied diseases.

show abstract

Sequential Feature Selection for Classification

Cited by 62 publications

References 8 publications

Outcome Prediction for Patient with High-Grade Gliomas from Brain Functional and Structural Networks

Outcome Prediction for Patient with High-Grade Gliomas from Brain Functional and Structural Networks

Sequential approaches for learning datum-wise sparse representations

Novel Data Mining Methodology for Healthcare Applied to a New Model to Diagnose Metabolic Syndrome without a Blood Test

Contact Info

Product

Resources

About