The rise of social media has led to an increasing online cyber-war via hate and violent comments or speeches, and even slick videos that lead to the promotion of extremism and radicalization. An analysis to sense cyber-extreme content from microblogging sites, specifically Twitter, is a challenging, and an evolving research area since it poses several challenges owing short, noisy, context-dependent, and dynamic nature content. The related tweets were crawled using query words and then carefully labelled into two classes: Extreme (having two sub-classes: pro-Afghanistan government and pro-Taliban) and Neutral. An Exploratory Data Analysis (EDA) using Principal Component Analysis (PCA), was performed for tweets data (having Term Frequency—Inverse Document Frequency (TF-IDF) features) to reduce a high-dimensional data space into a low-dimensional (usually 2-D or 3-D) space. PCA-based visualization has shown better cluster separation between two classes (extreme and neutral), whereas cluster separation, within sub-classes of extreme class, was not clear. The paper also discusses the pros and cons of applying PCA as an EDA in the context of textual data that is usually represented by a high-dimensional feature set. Furthermore, the classification algorithms like naïve Bayes’, K Nearest Neighbors (KNN), random forest, Support Vector Machine (SVM) and ensemble classification methods (with bagging and boosting), etc., were applied with PCA-based reduced features and with a complete set of features (TF-IDF features extracted from n-gram terms in the tweets). The analysis has shown that an SVM demonstrated an average accuracy of 84% compared with other classification models. It is pertinent to mention that this is the novel reported research work in the context of Afghanistan war zone for Twitter content analysis using machine learning methods.

Currently, law enforcement and legal consultants are heavily utilizing social media platforms to easily access data associated with the preparators of illegitimate events. However, accessing this publicly available information for legal use is technically challenging and legally intricate due to heterogeneous and unstructured data and privacy laws, thus generating massive workloads of cognitively demanding cases for investigators. Therefore, it is critical to develop solutions and tools that can assist investigators in their work and decision making. Automating digital forensics is not exclusively a technical problem; the technical issues are always coupled with privacy and legal matters. Here, we introduce a multi-layer automation approach that addresses the automation issues from collection to evidence analysis in online social network forensics. Finally, we propose a set of analysis operators based on domain correlations. These operators can be embedded in software tools to help the investigators draw realistic conclusions. These operators are implemented using Twitter ontology and tested through a case study. This study describes a proof-of-concept approach for forensic automation on online social networks.

In this manuscript, variants of Jacobi solver implementation on general purpose graphical processing units (GPGPU) have been purposed and compared. During this work, parallel implementation of finite element method (FEM) using Poisson’s equation on shared memory architecture as well as on GPGPUs has been observed to identify computationally most expensive part of FEM software, which is linear algebra Jacobi solver. Sparse matrices were used for system of linear equations. Nine implementations of Jacobi solver have been developed and compared using various synchronization and computation methods like atomicAdd, atomicAdd_block, butterfly communication, grid synchronization, hybrid and whole GPU based computation methods, respectively. Experiments have showed that Jacobi implementations based on our implemented Butterfly communication method have outperformed CUDA 10.0 provided critical execution methods like atomicAdd, atomicAdd_block and grid methods. The GPU has achieved a max speedup of 46 times using GTX 1060 and 60 times using Quadro P4000 with double precision computations when compared with sequential implementation on Core-i7 8750H. All the developments were performed using C/C++ GNU compiler 7.3.0 on Ubuntu 18.04 and CUDA 10.0.

FEM solver is designed for shared memory architectures using an object-oriented paradigm. The FEM solver is divided into five major sections; these are a collection of mesh data, creating an object for each finite element, application of boundary conditions, assembly of a linear system of equations, and lastly a solver to calculate the solution for the system of linear equations. This work has focused on the study of multiple parallel implementations for each stage of the FEM solver. All the stages are implemented in C++ while each stage parallel regions are implemented using OpenMP. The data structures used in this work are also C++ objects implemented using STL containers, like sparse matrix used is constructed as two-layer nested binary trees where each internal tree represents a sparse row in a global matrix. These structures are used in every stage and have never been converted into more conventional matrix structures like CSR, etc. Finally, this work has shown the contribution of time for each stage and the amount of speedup achieved for them. Using 16 threads, the solver has attained a speedup of 5.21 for mesh having 2099201 unknowns and Poissons equation. Two linear solvers Jacobi and CG are used which were consuming 95 of total solver time. The novelty of our work lies in the fact that this work focused on the implementation of Finite Element Application (FEA) using object-oriented programming for a shared memory architecture using OpenMP without considering conventional techniques of application speedup and specialized data structures like CSR, etc.

Accurate fruit counting is one of the important phenotypic traits for crucial fruit harvesting decision making. Existing approaches perform counting through detection or regression counting. Detection of fruit instances is very challenging because of the very small fruit size as compared to the whole size image of a tree, while regression-based counting gives impressive results but becomes inaccurate when the number of instances increases. Moreover, most approaches lack scalability and are applicable only on one or two fruit types. In this paper, we propose a fruit counting mechanism that combines loose segmentation and regression counting that works on six fruit types, such as Apple, Orange, Tomato, Peach, Pomegranate and Almond. Through relaxed segmentation, fruit clusters are segmented to extract the small image regions which contain the small cluster of fruits. Extracted regions are forwarded for the regression counting of fruits. Relaxed segmentation is achieved through a state-of-the-art deconvolutional network, while modified Inception Residual Networks (ResNet) based nonlinear regression module is proposed for fruit counting. For segmentation, 4,820 original images, including corresponding mask images, of all six fruit types are augmented to 32,412 images through different augmentation techniques, while 21,450 extracted patches are augmented to 89,120 images used for the regression module training. The proposed approach has superseded the counting accuracy of existing approaches of individual fruit types, but we have achieved an overall 94.71% accuracy.