Drug Usage Safety from Drug Reviews with Hybrid Machine Learning Approach

With the increasing usage of drugs to remedy different diseases, drug safety has become crucial over the past few years. Often medicine from several companies is offered for a single disease that involves the same/similar substances with slightly different formulae. Such diversification is both helpful and dangerous as such medicine proves to be more effective or shows side effects to different patients. Despite clinical trials, side effects are reported when the medicine is used by the mass public, of which several such experiences are shared on social media platforms. A system capable of analyzing such reviews could be very helpful to assist healthcare professionals and companies for evaluating the safety of drugs after it has been marketed. Sentiment analysis of drug reviews has a large potential for providing valuable insights into these cases. Therefore, this study proposes an approach to perform analysis on the drug safety reviews using lexicon-based and deep learning techniques. A dataset acquired from the ‘Drugs.Com’ containing reviews of drug-related side effects and reactions, is used for experiments. A lexicon-based approach, Textblob is used to extract the positive, negative or neutral sentiment from the review text. Review classification is achieved using a novel hybrid deep learning model of convolutional neural networks and long short-term memory (CNN-LSTM) network. The CNN is used at the first level to extract the appropriate features while LSTM is used at the second level. Several well-known machine learning models including logistic regression, random forest, decision tree, and AdaBoost are evaluated using term frequency-inverse document frequency (TF-IDF), a bag of words (BoW), feature union of (TF-IDF + BoW), and lexicon-based methods. Performance analysis with machine learning models, long short term memory and convolutional neural network models, and state-of-the-art approaches indicate that the proposed CNN-LSTM model shows superior performance with an 0.96 accuracy. We also performed a statistical significance T-test to show the significance of the proposed CNN-LSTM model in comparison with other approaches.     

A supply network-modelling system for a small- to medium-sized manufacturing company

During the late 1990s and early 2000s, the profile of global manufacturing has experienced many changes. There is anecdotal evidence that many western manufacturing companies have chosen to expand their manufacturing base across geographical boundaries. The common reasons sited for these ventures are to exploit less expensive labour markets, to establish a presence in expanding markets and in response to the threat of new competition. Whilst a global manufacturing base can prove to have many cost and sales benefits, there are also many disadvantages. Logistics operations can often increase in complexity leading to higher reliance on planning and effective interpretation of demand data. In response, systems modelling has remerged as a fertile research area after many years. Many modelling and simulation techniques have been developed, but these have had very limited practical success. The authors have identified that majority of these simulation techniques rely upon a detailed market structure being known, when this is rarely the case. This paper describes the outcome of a research project to develop of a pragmatic set of tools to gather, assess and verify supply chain structure data. A hybrid collection of technologies are utilised to assist these operations and to build a dynamic supply network model.     

Hardware/software co-design of a real-time kernel based tracking system

The probabilistic visual tracking methods using color histograms have been proven to be robust to target model variations and background illumination changes as shown by the recent research. However, the required computational cost is high due to intensive image data processing. The embedded solution of such algorithms become challenging due to high computational power demand and algorithm complexity. This paper presents a hardware/software co-design architecture for implementation of the well-known kernel based mean shift tracking algorithm. The design uses color histogram of the target as tracking feature. The target is searched in the consecutive images by maximizing the statistical match of the color distributions. The target localization is based on gradient based iterative search instead of exhaustive search which makes the system capable of achieving frame rate up to hundreds of frames per second while tracking multiple targets. The design, which is fully standalone, is implemented on a low-cost medium-size field programmable gate array (FPGA) device. The hardware cost of the design is compared with some other tracking systems. The performance of the system in terms of speed is evaluated and compared with the software based implementation. It is expected that the proposed solution will find its utility in applications like embedded automatic video surveillance systems.     

A secure biometric based multi-server authentication scheme for social multimedia networks

Social networking is one of the major source of massive data. Such data is not only difficult to store, manipulate and maintain but it’s open access makes it security prone. Therefore, robust and efficient authentication should be devised to make it invincible against the known security attacks. Moreover, social networking services are intrinsically multi-server environments, therefore compatible and suitable authentication should be designed accordingly. Sundry authentication protocols are being utilized at the moment and many of them are designed for single server architecture. This type of remote architecture resists each user to get itself register with each server if multiple servers are employed to offer online social services. Recently multi-server architecture for authentication has replaced the single server architecture, and it enable users to register once and procure services from multiple servers. A short time ago, Lu et al. presented two authentication schemes based on three factors. Furthermore, both Lu et al.’s schemes are designed for multi-server architecture. Lu et al. claimed the schemes to be invincible against the known attacks. However, this paper shows that one of the Lu et al.’s scheme is susceptible to user anonymity violation and impersonation attacks, whereas Lu et al.’s second scheme is susceptible to user impersonation attack. Therefore an enhanced scheme is introduced in this paper. The proposed scheme is more robust than subsisting schemes. The proposed scheme is thoroughly verified and validated with formal and informal security discussion, and through the popular automated tool ProVerif. The in-depth analysis affirms that proposed scheme is lightweight in terms of computations while attaining mutual authentication and is invincible against the known attacks, hence is more suitable for automated big data analysis for social multimedia networking environments.     

An efficient and robust integration technique for applied random vibration analysis

The mode-based finite element formulation of the equations of motion is usually adopted for linear random vibration analysis (RVA). In general, the RVA of large systems requires a large number of numerical integrations which is very time-consuming for a reasonable level of desired accuracy. Moreover, conventional numerical integration methods may fail to converge when the integrands are highly oscillatory due to slow propagation velocities. In this paper, a robust general-purpose RVA integration technique which can overcome these drawbacks is presented. Multi-point base and nodal excitations including wave passage effect and frequency-independent spatial correlation can be taken into account in the analysis. The proposed technique is based on the closed-form solutions for polynomial-type power spectral density functions and has been verified to be efficient and accurate for many engineering problems. This paper describes the implementation details, presents two examples taken from engineering applications and demonstrates the dramatic time-saving in the computation compared to numerical integration solutions. Response statistics, such as standard deviation of structural responses, are computed and displayed over the entire structures for these examples.     

Mixed formulation of nonlinear beam on foundation elements

This paper presents an inelastic element for the analysis of beams on foundations. The element is derived from a two-field mixed formulation with independent approximation of forces and displacements. The state determination algorithm for the implementation of the element in a general purpose nonlinear finite element analysis program is presented and its stability characteristics are discussed. Numerical studies are performed to compare the model with the classical displacement formulation. The studies confirm the superiority of the proposed model in describing the inelastic behavior of beams on foundations.     

Exploring the fitness landscape and the run-time behaviour of an iterated local search algorithm for cost-based abduction

Cost-based abduction (CBA) is an important problem in reasoning under uncertainty, and can be considered a generalization of belief revision. CBA is known to be NP-hard and has been a subject of considerable research over the past decade. In this paper, we investigate the fitness landscape for CBA, by looking at fitness–distance correlation for local minima and at landscape ruggedness. Our results indicate that stochastic local search techniques would be promising on this problem. We go on to present an iterated local search algorithm based on hill-climbing, tabu search, and simulated annealing. We compare the performance of our algorithm to simulated annealing, and to Santos' integer linear programming method for CBA.     

The space–time CESE scheme for shallow water equations incorporating variable bottom topography and horizontal temperature gradients

The effects of bottom topography and horizontal temperature gradients on the shallow water flows are theoretically investigated. The considered systems of partial differential equations (PDEs) are non-strictly hyperbolic and non-conservative due to the presence of non-conservative differential terms on the right hand side. The solutions of these model equations are very challenging for a numerical scheme. Thus, our primary goal is to introduce an improved numerical scheme which can handle the non-conservative differential terms efficiently and accurately. In this paper, the space–time conservation element and solution element (CESE) method is extended to approximate these model equations. The proposed scheme has capability to overcome all difficulties posed by this nonlinear system of PDEs. The performance of the scheme is analyzed by considering several case studies of practical interest and the results of suggested scheme are compared with those of central NT scheme. The accuracy of the scheme is verified qualitatively and quantitatively.     

SUPPORT VECTOR MACHINES (SVMs) WITH UNIVERSAL KERNELS

In this paper, a universal kernel function is introduced that could improve the classification accuracy of Support Vector Machines (SVMs) for both linear and nonlinear data sets. A class of universal kernel functions on the basis of the properties of the common kernels is proposed, which can find numerous applications in practice. The proposed kernels satisfy Mercer's condition and can be used for generating the most established kernels such as Gaussian Radial Function (GRF), Polynomial Radial Basis Function (PRBF), and Polynomial Exponential Radial Function (PERF) of SVMs. The SVM with the universal kernel is experimentally applied to a variety of nonseparable data sets with several attributes, leading to good classification accuracy in nearly all the data sets, especially those of high dimensions. The use of the universal kernel results in a better performance than those with established kernels.     

Fixed frequency sliding mode control of renewable energy resources in DC micro grid

The rising cost of fossil fuels, their high depleting rate and issues regarding environmental pollution have brought the attention of the researchers towards renewable energy technologies. Different renewable energy resources like wind turbines, fuel cells and solar cells are connected to DC micro grid through controllable power electronic converters. In presence of these diverse generation units, robust controllers are required to ensure good power quality and to regulate grid voltage. This paper presents a sliding mode control based methodology to address the above mentioned challenges. The proposed technique keeps the switching frequency constant so that electromagnetic compatibility (EMC) issues can be solved with conventional filter design. Parallel operation of converter in DC micro gird is considered. Chattering reduction and power quality improvement by harmonic cancellation is proposed. A scaled down hardware for unregulated 11.5 V to 17.5 V input and 24V output is designed and tested. The experimental results show good performance of the controller under different loads and uncertain input voltage conditions. Moreover, the results show the robustness of the closed loop system to sudden variations in load conditions. Furthermore, a significant improvement in power quality is achieved by harmonic cancellation of chattering in the output of the converters.