This paper focuses on analyzing the text sentiment tendency based on the deep learning model and starts with improving the neural network model based on public corpora to provide fine-grained analysis of text sentiment tendency and more accurate predictions. In the existing research, the extraction and utilization of text emotional features are usually based on Valence-Arousal space (VA space), but they do not pay attention to some subjective text details with emotional tendencies, such as the punctuation marks or emotional words, which could in turn to decrease the prediction accuracies made by models. Aiming at this issue, this paper proposes a hybrid Bidirectional Long Short-term Memory (BiLSTM) and Convolution Neural Network (CNN) model with an attention mechanism. Notably, in order to make our models easier to be applied to some light-weight products, we adopt the most basic components of nature language process (NLP) models. Firstly, BiLSTM is used to extract bidirectional context dependency information, and an attention mechanism is exploited to assign different weights to words that play different roles in sentiment judgment. Further, CNN is used to extract the local features of the upper layer’s output to ensure the robustness of feature extraction. The experiment shows that the combination of the methods with this order, BiLSTM first and CNN later, can achieve results which are significantly better than the existing baseline models reported in the literature.

Load balancing has an important role in grid computing for managing jobs and allocating them to adequate resources. An efficient load balancing algorithm can minimize execution time and maximize resource utilization by achieving the load balance between resources in the grid. In this paper, a simple and efficient fuzzy-based load-balancing algorithm is proposed that not only offers efficient resource management but also ensures high utilization of dynamic resources. In this paper, we detail the design and implementation of a proposed fuzzy-logic-based load balancing algorithm in grid computing. The proposed algorithm works by using a fuzzy logic inference system that uses some metrics to capture the load of CPU and the queue length for specifying the state of each node. Then, based on the overall nodes' states, the state of the corresponding cluster will be defined to allocate the newly arrived jobs such that load balancing among different clusters and nodes is performed. The performance of the proposed algorithm is evaluated based on resource utilization, response time, and tardiness, and it is compared with scheduling algorithms like First Come First Serve (FCFS), SJF(Shortest Job First), Earliest Deadline First (EDF), and Job Migration Algorithm for Dynamic Load Balancing (JMADLB). To analyze the performance of fuzzy based load balancing algorithm, a simulation is carried out using two Java library jFuzzyLogic and GridSim simulator. The results are presented in detail.

With the rapid growth of cloud or on-premises storage, widespread networking and other physical devices, complex IT infrastructure and processes for creating, processing, and storing all forms of electronic data, securing data that too in the initial compromise phase is critical, so it doesn’t translate back into a cyber-attack. During the covid pandemic where the whole world was working from open networks, data breaches and cyber security issues have gone to their peak. Intruders are moving laterally compromising data intelligently following various techniques like delaying the attack cycle where the intruder enters the network, gathers information, and stays away for a couple of days so that their previous interaction goes faded. This article describes the collaborative pattern analysis and event chaining of the compromised data during the reconnaissance phase of the attack chain and enables deception technology by enhancing predictability and planning to place decoys in the network dynamically. Deception technology can understand instantaneous data and provide verdicts based on real-time interactions. If any suspicious behaviour encountered by the decoys for that instance is co-related well with the attack matrix an alarm is being raised where decoys generate lures which create a false information stream that leads attackers exposed while protecting real enterprise network and assets. However, the current setup is not enabled to a level where it can track attackers who are working from multiple endpoints at the same time or using the gained data from the scans to access in the future. We are proposing a new and efficient Event chaining-based solution (named as DT-Chains) that overcomes the limitations in earlier proposed solutions. As part of this framework, we propose to design and develop a solution that will do an analysis on reconnaissance Attack Data. This newly proposed solution is expected to enable existing deception Technology to reduce false positives and helps to track attackers which are working from multiple endpoints at the same time or using the gained data from the scans to access in the future. This will also help in predicting the attack critical path which enables automated deception triggers for decoys.

As a consequence of fierce global competition and a shorter product life cycle, sometimes demand from the market exceeds the manufacturing capacity of the organization. Under such situations constrained resources limit production capacity through the first part of this paper, an attempt has been made to find out the optimal product mix in case of multiple constraints by using the Theory of Constraints (TOC) and Linear-Integer Programming (LIP). In this case of multiple resource constraints, the throughput generated by TOC is more than that of LIP, but the solution, provided by TOC, was found infeasible. Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) is used for to recital assessment of investment alternatives with a view to exploring the opportunity embedded in multiple resource constraint problems. Capital investment is a major decisive factor in any investment process and it has to be separately emphasized. In the final section an algorithm, holistic in nature, is proposed to find out the best alternative in a utopian environment. The proposed algorithm is a hybrid of TOPSIS and capital investment. To make an eclectic decision sensitivity analysis was also reported.

Much research has been done in the fields of argument mining, persuasiveness and sentiment analysis. Argument mining is a compound operation that encapsulates a lot of processing to reach its analysis state. An argument first must be mined; then, it has to be broken into the claim/s and the premise/s using argument structure identification and the relationship prediction between argument’s various structural aspects. Further, the context of argument is determined using the key aspects of argument. Persuasiveness is identification of influential power that is there in a given input. Sentiment Analysis rotates around identification of emotions and their corresponding intensities. We have attempted to amalgamate all these three 'Argument', 'Sentiment' and 'Persuasiveness' to form an argument persuasiveness prediction model based on sentiment analysis. This paper brings to table a unique approach to solve the classification problem of classifying whether the argument is persuasive or not based on sentiment analysis empirically.

In technical terms, an image is the visual representation of a thing or person created using optical technology (such as a mirror or lens) or a technological apparatus. 2 distinct domains of multiresolution image fusion, i.e., PCA and CNN, are discussed in this study. Multiresolution image fusion techniques amalgamate more than two images covering optical unclear and blurred parts to produce an image covering all the focused areas or information. Based on the study, in both PCA and CNN. PCA is more straightforward among all image fusion approaches; meanwhile, according to the study conducted in this paper, it produces less effective results. On the other hand, CNN gives more effective results, but it is complex to handle. Also, the boundary pixels of the fused image has some mismatching problems, i.e., unrecognizable pixels. The effectiveness of the results is measured based on some statistical image quality parameters.

To prioritize treatment, allocate resources efficiently, and reduce negative consequences, identifying clinically unstable patients as early as possible is crucial. Consistent indicators of failure and death include cardiorespiratory instability and sepsis; hence, they are often used as the foundation for aggregate-weighted, vital sign-based early warning systems. Recently, it has been shown that aggregate-weighted models fall short, although machine learning models have shown promise due to their capacity to include trends and capture parameter correlations. Our primary objective in conducting this research would be to discover, examine, and evaluate the available literature on machine learning-based system for early warning that utilize vital signs to identify the likelihood of decline of one's physical condition in critically ill inpatients and outpatients. Several electronic databases were searched using the phrases "vital signs," "clinical deterioration," and "machine learning" to locate research papers. These studies all describe a machine learning model that uses demographic and vital sign data to predict emergency and outpatient outcomes. Throughout the data extraction procedure, the SRMA, TRMPMIPD, and Cochrane Collaboration procedures were observed.

In this paper, we present a Markov reward model for a linear consecutive 2-out-of-4:F system with multi-states assuming different performance ranges from working state to failure state. Reliability measures for the suggested system such as average availability, the mean number of system failures, mean time to failure, and system reliability are obtained.