Semisupervised Feature Selection Based on Relevance and Redundancy Criteria

Xu, Jin; Tang, Bo; He, Haibo; Man, Hong

doi:10.1109/tnnls.2016.2562670

Cited by 134 publications

(55 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Figure 1a shows the data set in the first two dimensions, in which two small Gaussian clusters are buried in one red class. We compared SADA with five methods, including sSelect (Zhao and Liu 2007), LSDF (Zhao, Lu, and H 2008), PRPC (Xu et al 2016), RLSR (Chen et al 2017) and DSFFS (Yuan et al 2018). In this experiment, the projection dimension was set as 1 and the nearest neighborhoods k was set as 5.…”

Section: Experimental Results and Analysismentioning

confidence: 99%

Semi-Supervised Feature Selection with Adaptive Discriminant Analysis

Zhong

Chen

Yuan

et al. 2019

AAAI

View full text Add to dashboard Cite

In this paper, we propose a novel Adaptive Discriminant Analysis for semi-supervised feature selection, namely SADA. Instead of computing fixed similarities before performing feature selection, SADA simultaneously learns an adaptive similarity matrix S and a projection matrix W with an iterative method. In each iteration, S is computed from the projected distance with the learned W and W is computed with the learned S. Therefore, SADA can learn better projection matrix W by weakening the effect of noise features with the adaptive similarity matrix. Experimental results on 4 data sets show the superiority of SADA compared to 5 semisupervised feature selection methods.

show abstract

Section: Experimental Results and Analysismentioning

confidence: 99%

Semi-Supervised Feature Selection with Adaptive Discriminant Analysis

Zhong

Chen

Yuan

et al. 2019

AAAI

View full text Add to dashboard Cite

show abstract

“…As a filter feature selection method, Pearson correlation coefficient has strong generality and low complexity, and it has strong advantages in dealing with large-scale data sets and can eliminate a large number of irrelevant features in a short time. Therefore, it is often used for feature selection of the whole data set [34]. The formulation of Pearson correlation coefficient is represented in Equation 6:…”

Section: Proposed Emd-gru Model With Feature Selectionmentioning

confidence: 99%

Short-Term Electricity Load Forecasting Model Based on EMD-GRU with Feature Selection

Gao

Zhao

et al. 2019

Energies

View full text Add to dashboard Cite

Many factors affect short-term electric load, and the superposition of these factors leads to it being non-linear and non-stationary. Separating different load components from the original load series can help to improve the accuracy of prediction, but the direct modeling and predicting of the decomposed time series components will give rise to multiple random errors and increase the workload of prediction. This paper proposes a short-term electricity load forecasting model based on an empirical mode decomposition-gated recurrent unit (EMD-GRU) with feature selection (FS-EMD-GRU). First, the original load series is decomposed into several sub-series by EMD. Then, we analyze the correlation between the sub-series and the original load series through the Pearson correlation coefficient method. Some sub-series with high correlation with the original load series are selected as features and input into the GRU network together with the original load series to establish the prediction model. Three public data sets provided by the U.S. public utility and the load data from a region in northwestern China were used to evaluate the effectiveness of the proposed method. The experiment results showed that the average prediction accuracy of the proposed method on four data sets was 96.9%, 95.31%, 95.72%, and 97.17% respectively. Compared to a single GRU, support vector regression (SVR), random forest (RF) models and EMD-GRU, EMD-SVR, EMD-RF models, the prediction accuracy of the proposed method in this paper was higher.

show abstract

“…Feature selection aims to gain relevant features and remove redundant features for reducing computational cost,() which can effectively avoid traveling all attributes. The max‐relevance and min‐redundancy criterion based on Pearson's correlation (RRPC) coefficient is an effective method to select optimal feature subsets. The RRPC algorithm achieves a good balance between relevance and redundancy in feature selection.…”

Section: A Fast Rank Mutual Information Based Decision Treementioning

confidence: 99%

“…Suppose the training data X with n condition attributes

A = {false{A_{k} false}}_{k = 1}^{n}

and one decision attribute D can be split into m subsets

{false{X_{i} false}}_{i = 1}^{m}

in the framework of Map‐Reduce. In the Map phase, we first apply RRPC algorithm to generate a subset A ′ of A from X i . Then, we compute the best splitting point

c p_{j}^{*} false(i false)

for each attribute

A_{k^{'}}

in A ′ by the following formula:

c p_{j}^{*} false(i false) = \arg \max_{c p_{j}} R M I_{c p_{j}} false(A_{k^{'}} false),

where

R M I_{c p_{j}} false(A_{k^{'}} false) = R M I^{\leq} false(A_{k^{'}}, c p_{j}, D false)

is calculated by Equation and c p j is a center of A k generated by FCM algorithm.…”

Section: The Parallel Fast Rank Mutual Information Based Decision Treementioning

confidence: 99%

A fast rank mutual information based decision tree and its implementation via Map‐Reduce

Wang

Liu

2018

Concurrency and Computation

View full text Add to dashboard Cite

Summary To address the time‐consuming problem for the confirmation of splitting attributes and splitting points in classic rank mutual information based decision trees, this paper establishes a fast rank mutual information based decision tree (FRMIDT) for classification problems. First, the proposed FRMIDT algorithm improves the velocity by a max‐relevance and min‐redundancy criterion to remove the redundant attributes in each tree node building. Then, the fuzzy c‐means algorithm is employed to confirm the splitting points for further acceleration. Meanwhile, a parallel implementation is developed in the framework of Map‐Reduce (MR‐FRMIDT) for medium or large‐scale data classification. Several comparative studies are conducted on UCI benchmark data sets. In contrast to the classic rank mutual information based decision tree on 12 data sets, the proposed FRMIDT model effectively reduces the computational time on the premise of keeping testing accuracy. Furthermore, the proposed FRMIDT algorithm is comparable through comparing FRMIDT with other traditional decision tree classifiers including BFT, C4.5, LAD, NBT, and SC. Meanwhile, the comparison with 7 different popular splitting measures based monotonic decision trees on several data sets illustrates the effectiveness of FRMIDT in monotonic classification. At last, the experimental analysis on other 6 data sets shows that the proposed MR‐FRMIDT is feasible and has a good parallel performance on reducing execution time and avoiding memory restrictions.

show abstract

Semisupervised Feature Selection Based on Relevance and Redundancy Criteria

Cited by 134 publications

References 47 publications

Semi-Supervised Feature Selection with Adaptive Discriminant Analysis

Semi-Supervised Feature Selection with Adaptive Discriminant Analysis

Short-Term Electricity Load Forecasting Model Based on EMD-GRU with Feature Selection

A fast rank mutual information based decision tree and its implementation via Map‐Reduce

Contact Info

Product

Resources

About