Transactions on Computational Science XXXIV

Multi sensor image data used in diverse applications for Earth observation has portrayed immense potential as a resourceful foundation of information in current context. The scenario has kindled the requirement for efficient content-based image identification from the archived image databases to provide increased insight to the remote sensing platform. Machine learning is the buzzword for contemporary data driven decision making in the domain of emerging trends in computer science. Diverse applications of machine learning have exhibited promising outcomes in recent times in the areas of autonomous vehicles, natural language processing, computer vision and web searching. An important application of machine learning is to extract meaningful signatures from the unstructured data. The process facilitates identification of important information in the hour of need. In this work, the authors have explored the application of machine learning for content based image classification with remotely sensed image data. A hybrid approach of machine learning is implemented in this work for enhancing the classification accuracy and to use classification as a pre cursor of retrieval. Further, the approaches are compared with respect to their classification performances. Observed results have revealed the superiority of the hybrid approach of classification over the individual classification results. The feature extraction techniques proposed in this work have ensured higher accuracy compared to state-of-the-art feature extraction techniques.

High-utility patterns generated from mining the unknown and different databases can be clustered to identify the most valid patterns. Sources include the internet, journals, and enterprise data. Here, a grid-based clustering method (CLIQUE) is used to aggregate patterns mined from multiple databases. The proposed model forms the clusters based on all the utilities of patterns to determine the interestingness and the correct interval of its utility measure. The set of all patterns is collected by first mining the databases individually, at the local level. The problem arises when the same pattern is identified by all of the databases but with different utility factors. In this case, it becomes difficult to decide whether the pattern should be considered as a valid or not, due to the presence of multiple utility values. Hence, an aggregation model is applied to test whether a pattern satisfies the utility threshold set by a domain expert. We found that the proposed aggregation model effectively clusters all of the interesting patterns by discarding those patterns that do not satisfy the threshold condition. The proposed model accurately optimizes the utility interval of the valid patterns.

The point placement problem is to determine the locations of n distinct points on a line uniquely (up to translation and reflection) by making the fewest possible pairwise distance queries of an adversary (an adversary is just a source of true distances). A number of deterministic and randomized algorithms are available when distances are known exactly. In this paper, we discuss the problem in an inexact model. This is when distances returned by the adversary are not exact; instead, only upper and lower bounds on the distances are provided. We explore three different approaches to this problem. The first is based on an adaption of a distance geometry approach that Havel and Crippen [6] used to solve the molecular conformation problem. The second is based on a linear programming solution to a set of difference constraints that was used by Mumey [7] to solve a probe location problem arsing in DNA sequence analysis. The third is based on a heuristic called Stochastic Proximity Embedding, proposed by Agrafiotis [8]. Extensive experiments were carried out to determine the most promising approach vis-a-vis two parameters: run-time and quality of the embedding, measured by computing a certain stress function.

Inspired by the recent growth of computational methods for general polygonal and polyhedral meshes, this paper introduces Cinolib: a novel header only C++ library for geometry processing. Cinolib differentiates itself from similar toolkits in that it is specifically designed to support a wide set of meshes, such as triangle, quadrilateral and general polygonal surface meshes, as well as tetrahedral, hexahedral and general polyhedral volumetric meshes. At the core of the library there is a hierarchical data structure that factorizes the common properties among the various meshes, allowing tools and algorithms to operate on the widest possible set of meshes with a single implementation, thus avoiding code repetition and facilitating bug fixing and software maintenance. Cinolib is licensed with MIT, it currently counts more than 50K lines of code and, besides the core structure, already comprises a vast set of widespread tools for computer graphics and engineering.

In this paper, we present a mechanism for trust computation in VANET cloud. The presence of VANET cloud allows vehicles to store their past trust values of other vehicles. These values are utilized for fast trust computation by other vehicles. The mechanism takes into consideration the uncertainty and fuzziness associate with trust values by incorporating DST (Dempster Shafer Theory) and fuzzy analyzer. The change in trust value due to action of a vehicle is done through a reward & penalty scheme. This proposed mechanism optimizes the execution and computation of a batch of simulations, increasing the overall performance, in terms of simulation time and costs. Simulation results indicate the DST based fuzzy trust mechanism is able to manage the trust of vehicles efficiently in the VANET cloud environment.

The work in this paper aims at showing the effectiveness of cooperative communication considering a Wireless Sensor Network (WSN) scenario deployed in a snowy environment. Sensor nodes with adjustable antenna heights are assumed so that even when the ground is covered with snow, these antenna nodes are able to transmit data. Received power of the deployed cooperative network with adjustable antenna heights has been evaluated and further using it as a crucial parameter a comparative analysis of cooperative and non-cooperative scenario is carried out. An Amplify and Forward cooperative protocol has been considered for modeling the network and deriving the expressions for energy and power. Further, various path loss models have been considered to compute the pathloss and evaluate the performance of network. Simulation and analytical results reveal that increasing the antenna height reduces the received power and cooperative scenario outperforms the non-cooperative scenario.

This paper presents a built-in self-reconfiguring system for mesh-connected processor arrays where faulty processing elements are compensated for by spare processing elements located on the diagonal. First, an algorithm for reconfiguring the arrays with faulty processing elements is presented. The reliability of the system is analyzed by simulation and compared with that of arrays having spare processing elements on one side. The result shows that under the condition of the same number of spares, the former is fairly higher than the latter. Next, a logical circuit realizing self-reconfiguration by the algorithm is described. The circuit controls interconnections of processing elements and its hardware overhead is shown to be quite small (i.e. less than ten logic gates for each processing element). The proposed system is effective for the case where each processing element is fairly reliable and the small number of spares is sufficient for retaining a high reliability. It is also effective in enhancing the run-time reliability of a processor array in mission critical systems where first self-reconfiguration is required without an external host computer or manual maintenance operations.