China
Africa
Explore chapters and articles related to this topic
Attacking Smartphone Security and Privacy
Published in Georgios Kambourakis, Asaf Shabtai, Constantinos Kolias, Dimitrios Damopoulos, Intrusion Detection and Prevention for Mobile Ecosystems, 2017
Vincent F. Taylor, Ivan Martinovic
iOS is a mobile operating system developed by Apple Inc. It has a healthy app ecosystem that surrounds it with over 1.4 million iOS applications available for download. The operating system itself is proprietary, closed source, and written in C, C++, Objective-C, and Swift. It is a Unix-like operating system and features a hybrid kernel that runs on 64/32-bit ARM processors. Before iOS apps are made available to the public in the Apple App Store, they must undergo a thorough vetting process by Apple. Apps must pass reliability testing and other analysis to ensure that they are not malicious or otherwise unsavory. Apple's vetting process includes manual testing and static analysis to determine whether an app tries to perform actions outside of what it claims to do [6]. This vetting process is not always perfect and indeed security researchers have uncovered ways of circumventing the protections put in place by Apple [7]. In the case of Jekyll [8], the malicious app passed the vetting process by rearranging its code to add new, malicious functionality, after passing the approval process. The iOS kernel uses code signing to ensure that all apps running on a device come from an approved source and have not been tampered with [9]. Additionally, all third-party apps are sandboxed by iOS to prevent them from accessing data stored by other apps and modifying the system. However, Han et al. described how to “break out” of the iOS sandbox by leveraging dynamically loaded, private APIs in malicious apps [10]. Finally, iOS enforces a secure boot chain and file encryption using a per-file key.
Optimizing the Estimation of Nonlinear Kernels
Published in Robert B. Pinter, Bahram Nabet, Nonlinear Vision: Determination of Neural Receptive Fields, Function, and Networks, 1992
Hybrid kernel normalization and summation conventions are used in Equation 81. The normalization is given by Equation 8: () Gn(τ,τ12⋯τp2) = gn(τ,τ12⋯τp2)n!/2p
Video-based Yogasan classification for the musculoskeletal disorder using the Cervus trail dependent multiclass support vector machine
Published in Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization, 2023
Suvarna Nandyal, Somashekhar S. Dhanyal
Proposed Hybrid Kernel Function designed for multiSVM classifier: The hybrid kernel function in the multiSVM classifier is utilised to manipulate the data, in which the transformation of the data occurs. Here, the optimum boundary is used to classify the Yogasan based on the features, which supports the classification decision-making. The kernels in the multiSVM are categorised into two different types, such as local and global kernels. In the global kernel, the points of the kernel are far away and in the local kernel, the kernel points remain near, for effective decision-making. The hybrid kernel function is designed using the kernel functions, such as sigmoid (S), radial basis function (R), Gaussian (G), Polynomial, and linear or straightforward kernel. The significance of sigmoid activation relies on its non-linear nature, which boosts the learning performance, but suffers from vanishing gradient, whereas the kernel functions, like Gaussian, RBF, and polynomial kernels enable high-dimensional mapping by forming a non-linear combination of features to support the linear classification of the samples. Hence, a hybrid kernel function is designed that supports the accurate classification using multiSVM classifier. The expression for the hybrid kernel function in multiSVM is formulated as,
MKELM: Mixed Kernel Extreme Learning Machine using BMDA optimization for web services based heart disease prediction in smart healthcare
Published in Computer Methods in Biomechanics and Biomedical Engineering, 2022
Adlin Sheeba, S. Padmakala, C. A. Subasini, S. P. Karuppiah
In mixed kernel based extreme learning machine (MKELM) learning algorithm, the regulation coefficient, as well as the kernel factors, is selected properly for enhancing the generalization neural networks performances. The hybrid kernel function is utilized for enhancing the MKELM performance. On the other hand, choosing the optimal value of the kernel function factors is not resolved and hence the Biogeography-based optimization algorithm and Mexican hat wavelet to enhance Dragonfly algorithm (BMDA) optimization algorithm are established along with the MKELM for selecting the optimal factors of the kernel function (Ali, Prasad et al. 2020). There are various optimization algorithms in the machine learning domain and compared with other algorithms, the Biogeography-based optimization algorithm and Mexican hat wavelet to enhance Dragonfly algorithm (BMDA) optimization method is the biologically inspired optimization technique. The BMDA algorithm is becoming famous due to its requirement of less memory, easiness in the implementation and capability for converging the best solution rapidly.
Intelligent travelling visitor estimation model with big data mining
Published in Enterprise Information Systems, 2021
He-Qing Zhang, Zhaojun (Steven) Li
Recently, a visitor estimation algorithm based on FOA Optimized Hybrid Kernel LSSVM was proposed (Geng and Chen 2017). It used the mixed kernel function of polynomial kernel and radial basis kernel as the LSSVM kernel function to construct a hybrid kernel LSSVM prediction model for railway freight volume, which can be generated to passenger volume estimation easily. At the same time, using the advantage of the FOA’s global optimisation ability and fast calculation speed, the mixed kernel LSSVM parameters are optimised and the number of visitors is estimated. However, during the estimation process of this algorithm, analysis of the factors affecting the estimation results was ignored, which leads to a low accuracy of the estimation results.