A Novel Fuzzy Measure and its Choquet Integral Regression Model

Liu, Hsiang-Chuan; Jheng, Yu-Du; Lin, Wen-Chih; Chen, Guey-Shya

doi:10.1109/icmlc.2007.4370362

Cited by 15 publications

(4 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There are different classes of fuzzy measures; they include probability (a subset of classical measures), plausibility, necessity, possibility, and belief measures. The following equations mark the method of determining a fuzzy measure [23], [24]:…”

Section: A Fuzzy Measurementioning

confidence: 99%

A New Fuzzy-DNA Image Encryption and Steganography Technique

2020

View full text Add to dashboard Cite

Image encryption and steganography techniques are receiving a lot of interest and investigations due to their high importance in multimedia communication systems. A novel highly efficient image encryption and steganography technique are presented in this paper. For the first time, the proposed technique uses hybrid DNA encoding and Choquet's Fuzzy Integral sequences. At first, a confused version of the image, using a simple chaotic map, is encoded using DNA's bases. Four coded images are generated using the four DNA bases, namely AT, CG, GC, and TA. Parallel to that, a Choquet's fuzzy Integral sequence is generated and DNA encoded similarly to obtain four pseudo-random sequences. Secondly, the resulting four fuzzy/DNA sequences are used to diffuse the four DNA encoded images using the complementary DNA XOR rule, according to certain control code. Finally, the wavelet fusion algorithm is then used to fuse the resulting four fuzzy-DNA encoded images, to get the encrypted image. For added security, a new steganography approach is used. In particular, the encrypted image is divided into four sub-images, each of which is hidden in a different carrier image selected from a known group of carrier images according to a given key. The simulation results and security analysis confirmed the efficiency of the proposed image encryption algorithm as well as the steganography approach used for enhanced security. Ten different images with a size of 256 × 256 are used to test the proposed method. The results show that the proposed algorithm has a higher key sensitivity. The pixel correlation coefficient values are very small (between 5.3220e-04 and 0.0011 horizontally, between 8.7670e-04 and 0.0022 vertically and between 0.0002 and 0.0045 diagonally). Furthermore, the measured information entropy of the encrypted images is between 7.9970 and 7.9979 which are very close to the ideal value of 8. Additionally, the measured unified average changing intensity and number of pixels change rate values take the values between 33.46 and 33.39 and between 99.61 and 99.64, respectively, which are again closed to the ideal values. The steganography test shows that the hidden encrypted images are almost invisible at high values of SNR and are characterized by good NCC values under different types of attacks. The performance of the new proposed algorithm is proved to overcomes many other previously published image encryption techniques.

show abstract

Section: A Fuzzy Measurementioning

confidence: 99%

A New Fuzzy-DNA Image Encryption and Steganography Technique

2020

View full text Add to dashboard Cite

show abstract

“…In recent years, a new fuzzy measure, L-measure, have been proposed by Liu et al [7]. When interactions among independent variables exist in forecasting problems, the Choquet integral based on the L-measure can improve this situation.…”

Section: Introductionmentioning

confidence: 99%

Combined multiple svm classifiers based on Choquet integral with respect to L- measure

Lin

Huang

2009

2009 International Conference on Machine Learning and Cybernetics

Self Cite

View full text Add to dashboard Cite

Combining multiple classifiers is a natural way to explore useful information and improve the performances of individual classifiers. Support vector machine (SVM) has an excellent ability to solve the classification problems. In this study, we try to combine the multiple SVMs which is desirous to gain a more accurate classification than single SVM. When interactions exist in combining multiple SVMs, fuzzy integral with respect to L-measure would be a valid method to fuse these multiple SVMs. From this experiment results, the fusion method based on this fuzzy fusion obtains advancement in terms of the performance of classification.

show abstract

“…Fuzzy measures [31,33,36,68] and fuzzy integrals [15,59,64,66,70] have been applied successfully in multi-attributes decision-making [18,53,54], classification [52,65,69], information fusion [3,5,11,43,49], nonlinear multi-regression [30], feature selection [19,44] and image processing [24,32,34,35]. The reason of success is from the highly non-additive and non-linear characteristics of fuzzy measures and fuzzy integrals.…”

Section: Introductionmentioning

confidence: 99%