Feature selection for face recognition using DCT-PCA and Bat algorithm

Preeti, .; Kumar, Dinesh

doi:10.1007/s41870-017-0051-6

Cited by 25 publications

(5 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…They have obtained a 97.5 % recognition rate with fewer features, which provides the best results compared to other models on the ORL dataset. They have applied their approach to three different datasets, YaleB Pose, YaleB Illumination, and YaleB, to compare their approach and generate better results [13].…”

Section: Related Workmentioning

confidence: 99%

Correlation Matrix as a Smart Filter for Malware Classification Using Ensemble of Novel Feature Selection Algorithms

Cürebal,

Demirkıran,

Çayır

et al. 2023

Preprint

View full text Add to dashboard Cite

While the development of technology has made our lives easier, our dependence on it has also increased. Cybercriminals develop various types of malware to exploit this dependency. Thus, study of malware, for example, malware classification, is essential for security researchers and incident response teams to take the necessary action against it to prevent or reduce the damage intended. This paper investigates the importance of features in a given malware dataset. To this end, we combine three recently proposed feature selection methods: Leave One Feature Out (LOFO) Importance, Feature Relevance-based Unsupervised Feature Selection (FRUFS), and A General Framework for Auto-Weighted Feature Selection via Global Redundancy Minimization (AGRM) to find the most effective subset of features for the objective at hand. This preprocessing is important as it allows working with fewer features, reducing the dimension of the problem at hand. We use the feature subsets from these three different feature selection methods with three different models: Extreme Gradient Boosting (XGBoost), Random Forest (RF), and Histogram-Based Gradient Boosting (HGB) classifiers using a stacking ensemble. From the nine prediction probabilities obtained, we eliminated the prediction probabilities containing similar information by setting a threshold in the correlation matrix. By inputting the final prediction probabilities to the Support Vector Machine (SVM) meta-classifier, we aim to display how effective the proposed scheme could be. Our proposed model obtains an average of 1.2% better classification accuracy than the tested feature selection methods on one of the well-known and most widely used malware datasets (Microsoft Malware Prediction dataset).

show abstract

Section: Related Workmentioning

confidence: 99%

Correlation Matrix as a Smart Filter for Malware Classification Using Ensemble of Novel Feature Selection Algorithms

Cürebal,

Demirkıran,

Çayır

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…The quality of the selected features directly affects the final result. Recently, many new feature selection methods [1,3,23,24,31] have been proposed. Abualigah [1] proposes a new feature selection method using particle swarm optimization algorithm with a novel weighting scheme and a detailed dimension reduction technique.…”

Section: Feature Selectionmentioning

confidence: 99%

“…This method is used to obtain a new subset of more informative features with low-dimensional space. Preeti et al [31] make use of DCT-PCA combination to reduce the dimensionality and extract the features followed by Bat algorithm to yield a set of features that proves to be the best for face recognition under uncontrolled environment. Lu et al [23] proposes Dynamic Weighted DPA (DWDPA) to enhance the DP of the selected DCTCs without pre-masking window.…”

Section: Feature Selectionmentioning

confidence: 99%

Hierarchical saliency mapping for weakly supervised object localization based on class activation mapping

Cheng

Zeng

et al. 2020

Multimed Tools Appl

View full text Add to dashboard Cite

Weakly supervised object localization is a basic research in the field of computer vision. In this paper, a hierarchical saliency mapping network for object localization is proposed and designed to avoid missing detailed information of potential object. Based on the classical convolution network, we remove the fully connected part and add multiple information extraction branches. The network extracts information from convolution layers of different scales to generate Hierarchical Saliency Map. Hierarchical Saliency Maps that include Hierarchical-Class Activation Map and Hierarchical-Spatial Pyramid Saliency Map fuse deep-level features and low-level features to locate object. The datasets used for testing are Caltech-UCSD Birds 200, Caltech101 and ImageNet. Compared with Class Activation Map and Spatial Pyramid Saliency Map, the localization accuracy has been improved. This method can be used for fine-grained classification, object tracking and other fields.

show abstract

“…The results must be performed on other datasets also. Preeti et al [ 15 ] proposed the novel FR by utilizing combination of DCT+PCA and bat algorithm. The DCT+PCA combination is used for compression and extraction of features.…”

Section: Introductionmentioning

confidence: 99%

“…Better in light & rotation change The novelty is not enough to declare them as efficient descriptor and no global method is used LBP + HOG + AC-LBP [ 9 ] The concept of arithmetic coding and merger with LBP and HOG proves very impressive Testing has been performed on small benchmarks and no machine learning algorithm is used MB-LBPUH [ 10 ] Feature dimension is on lower side and integration with HOG yields much stronger descriptor The demerit in this work is that PCA is used for feature compression GBSBP + LPQ [ 11 ] Effective in various image transformations such as light, emotion, scale and pose. Novel methodology is introduced in GBSBP Evaluation is performed on small datasets FBLCM [ 12 ] FBLCM explicitly improve the performance in edges among white and black pixels of binary image The evaluation is conducted on small datasets Gaussian + Gabor + LBP [ 13 ] The method is invariant to light and noise changes Lack of global feature extraction and testing only on single dataset LOG + HOG [ 14 ] This method produces effective results on plastic surgery dataset Results are restricted only to single dataset DCT + PCA [ 15 ] Reduces the feature size and extracts the best features for classification by using bat algo Only two datasets are used for the evaluation Machine Learning [ 16 ] Various feature extraction techniques and machine learning approaches are evaluated The deep neural networks are not used for accuracy enhancement Geometric + Histogram [ 17 ] Th integration proves out as the effective met...…”

Section: Introductionmentioning

confidence: 99%

Improved local descriptor (ILD): a novel fusion method in face recognition

Karanwal

2023

Int. j. inf. tecnol.

View full text Add to dashboard Cite

Literature suggests that by fusing multiple features there is immense improvement in the recognition rates as compared to the recognition rates of single descriptor. This motivate researchers to develop more and more fused descriptors by joining multiple features. Inspiring from the literature work, the proposed work launch novel local descriptor so-called Improved Local Descriptor (ILD), by joining features of 4 local descriptors. These are LBP, ELBP, MBP and LPQ. LBP captures local details. ELBP capture robust features in horizontal and vertical directions (elliptically) by using 3 × 5 and 5 × 3 patches. MBP minimizes image noise by median comparison to all the pixels and LPQ quantize the frequency components for obtaining feature size. These essential merits of 4 descriptors are encapsulated in one framework in the form of histogram feature. PCA is used further for compression and SVMs and NN are used for classification. Results on ORL, GT and Faces94 confirms strength of ILD, which beats separately implemented descriptors and various benchmark methods.

show abstract

Feature selection for face recognition using DCT-PCA and Bat algorithm

Cited by 25 publications

References 14 publications

Correlation Matrix as a Smart Filter for Malware Classification Using Ensemble of Novel Feature Selection Algorithms

Correlation Matrix as a Smart Filter for Malware Classification Using Ensemble of Novel Feature Selection Algorithms

Hierarchical saliency mapping for weakly supervised object localization based on class activation mapping

Improved local descriptor (ILD): a novel fusion method in face recognition

Contact Info

Product

Resources

About