Towards a component-based system model to improve the quality of highly configurable systems.

Due to ever-evolving software developments processes, companies are motivated to develop desired quality products quickly and effectively. Industries are now focusing on the delivery of configurable systems to provide several services to a wide range of customers by making different configurations in a single largest system. Nowadays, component-based systems are highly demanded due to their capability of reusability and restructuring of existing components to develop new systems. Moreover, product line engineering is the major branch of the component-based system for developing a series of systems. Software product line engineering (SPLE) provides the ability to design several software modifications according to customer needs in a cost-effective manner. Researchers are trying to tailor the software product line (SPL) process that integrates agile development technologies to overcome the issues faced during the execution of the SPL process such as delay in product delivery, restriction to requirements change, and exhaustive initial planning. The selection of suitable components, the need for documentation, and tracing back the user requirements in the agile-integrated product line (APL) models still need to improve. Furthermore, configurable systems demand the selected features to be the least dependent. In this paper, a hybrid APL model, quality enhanced application product line engineering (QeAPLE) is proposed that provides support for highly configurable systems (HCS) by evaluating the dependency of features before making the final selection. It also has a documentation and requirement traceability function to ensure that the product meets the desired quality. Two-fold assessments are undertaken to validate the suggested model, with the proposed model being deployed on an active project. After that, we evaluated the proposed model performance and effectiveness using after implementing it in a real-world environment and compared the results with an existing method using statistical analysis. The results of the experimental study proofs that the proposed model is practically and statistically significant as compared to the existing method in terms of effectiveness and participants' performance. Hence, the statistical results of the comparative analysis show that the proposed model improved ease of understanding and adaptability, required effort, high-quality achievement, and version management are significant i.e., more the 50% as compared to the exiting method i.e., less than 50%. The proposed model offers to assist in the development of a highly configurable system that achieves the needed quality. Therefore, the proposed model manages the variation identification, versions control, components dependency for correct selection of components, and validation activities from domain engineering to application engineering.

One-Dimensional Multi-Scale Domain Adaptive Network for Bearing-Fault Diagnosis under Varying Working Conditions.

Data-driven bearing-fault diagnosis methods have become a research hotspot recently. These methods have to meet two premises: (1) the distributions of the data to be tested and the training data are the same; (2) there are a large number of high-quality labeled data. However, machines usually work under different working conditions in practice, which challenges these prerequisites due to the fact that the data distributions under different working conditions are different. In this paper, the one-dimensional Multi-Scale Domain Adaptive Network (1D-MSDAN) is proposed to address this issue. The 1D-MSDAN is a kind of deep transfer model, which uses both feature adaptation and classifier adaptation to guide the multi-scale convolutional neural network to perform bearing-fault diagnosis under varying working conditions. Feature adaptation is performed by both multi-scale feature adaptation and multi-level feature adaptation, which helps in finding domain-invariant features by minimizing the distribution discrepancy between different working conditions by using the Multi-kernel Maximum Mean Discrepancy (MK-MMD). Furthermore, classifier adaptation is performed by entropy minimization in the target domain to bridge the source classifier and target classifier to further eliminate domain discrepancy. The Case Western Reserve University (CWRU) bearing database is used to validate the proposed 1D-MSDAN. The experimental results show that the diagnostic accuracy for the 12 transfer tasks performed by 1D-MSDAN was superior to that of the mainstream transfer learning models for bearing-fault diagnosis under variable working conditions. In addition, the transfer learning performance of 1D-MSDAN for multi-target domain adaptation and real industrial scenarios was also verified.

Effective and Reliable Framework for Lung Nodules Detection from CT Scan Images.

Lung cancer is considered more serious among other prevailing cancer types. One of the reasons for it is that it is usually not diagnosed until it has spread and by that time it becomes very difficult to treat. Early detection of lung cancer can significantly increase the chances of survival of a cancer patient. An effective nodule detection system can play a key role in early detection of lung cancer thus increasing the chances of successful treatment. In this research work, we have proposed a novel classification framework for nodule classification. The framework consists of multiple phases that include image contrast enhancement, segmentation, optimal feature extraction, followed by employment of these features for training and testing of Support Vector Machine. We have empirically tested the efficacy of our technique by utilizing the well-known Lung Image Consortium Database (LIDC) dataset. The empirical results suggest that the technique is highly effective for reducing the false positive rates. We were able to receive an impressive sensitivity rate of 97.45%.

Resilience of epidemics for SIS model on networks.

Epidemic propagation on complex networks has been widely investigated, mostly with invariant parameters. However, the process of epidemic propagation is not always constant. Epidemics can be affected by various perturbations and may bounce back to its original state, which is considered resilient. Here, we study the resilience of epidemics on networks, by introducing a different infection rate λ2 during SIS (susceptible-infected-susceptible) epidemic propagation to model perturbations (control state), whereas the infection rate is λ1 in the rest of time. Noticing that when λ1 is below λc, there is no resilience in the SIS model. Through simulations and theoretical analysis, we find that even for λ2 < λc, epidemics eventually could bounce back if the control duration is below a threshold. This critical control time for epidemic resilience, i.e., cdmax, seems to be predicted by the diameter (d) of the underlying network, with the quantitative relation cdmax ∼ dα. Our findings can help to design a better mitigation strategy for epidemics.

Real-time extended interface automata for software testing cases generation.

Testing and verification of the interface between software components are particularly important due to the large number of complex interactions, which requires the traditional modeling languages to overcome the existing shortcomings in the aspects of temporal information description and software testing input controlling. This paper presents the real-time extended interface automata (RTEIA) which adds clearer and more detailed temporal information description by the application of time words. We also establish the input interface automaton for every input in order to solve the problems of input controlling and interface covering nimbly when applied in the software testing field. Detailed definitions of the RTEIA and the testing cases generation algorithm are provided in this paper. The feasibility and efficiency of this method have been verified in the testing of one real aircraft braking system.

Improved ant algorithms for software testing cases generation.

Existing ant colony optimization (ACO) for software testing cases generation is a very popular domain in software testing engineering. However, the traditional ACO has flaws, as early search pheromone is relatively scarce, search efficiency is low, search model is too simple, positive feedback mechanism is easy to produce the phenomenon of stagnation and precocity. This paper introduces improved ACO for software testing cases generation: improved local pheromone update strategy for ant colony optimization, improved pheromone volatilization coefficient for ant colony optimization (IPVACO), and improved the global path pheromone update strategy for ant colony optimization (IGPACO). At last, we put forward a comprehensive improved ant colony optimization (ACIACO), which is based on all the above three methods. The proposed technique will be compared with random algorithm (RND) and genetic algorithm (GA) in terms of both efficiency and coverage. The results indicate that the improved method can effectively improve the search efficiency, restrain precocity, promote case coverage, and reduce the number of iterations.