Implementing Self Organising Map to Organise the Unstructured Data

Surface reconstruction is significant in reverse engineering because it should present the correct surface with minimum error using the data available. It has become a challenging process when the data are in the unstructured type and the existing methods are still suffering from accuracy issues. The unstructured data will produce an incorrect surface because there is no connectivity information among the data. So, the unstructured data should undergo the organising process to obtain the correct shape. The Self Organising Map (SOM) has been extensively applied in previous works to solve surface reconstruction problems. However, the performance of the SOM models has remained uncertain. It can be evaluated and tested using different types of data sets. The objectives of this research are to examine the performance and to determine the weaknesses of SOM models. 2D SOM, 3D SOM, and Cube Kohonen (CK) SOM models are investigated and tested using three data sets in this research. As shown in the experimental results, the CKSOM model has proved to perform better because it can represent the correct closed surface with the lowest minimum error.

An Effective Control of Hello Process for Routing Protocol in MANETs

In the mobile ad hoc network (MANET) update of link connectivity is necessary to refresh the neighbor tables in data transfer. A existing hello process periodically exchanges the link connectivity information, which is not adequate for dynamic topology. Here, slow update of neighbour table entries causes link failures which affect performance parameter as packet drop, maximum delay, energy consumption, and reduced throughput. In the dynamic hello technique, new neighbour nodes and lost neighbour nodes are used to compute link change rate (LCR) and hello-interval/refresh rate (r). Exchange of link connectivity information at a fast rate consumes unnecessary bandwidth and energy. In MANET resource wastage can be controlled by avoiding the re-route discovery, frequent error notification, and local repair in the entire network. We are enhancing the existing hello process, which shows significant improvement in performance.

Inside Information: Systematic within-node connectivity changes observed across tasks or groups

Mapping the human connectome and understanding its relationship to brain function holds tremendous clinical potential. The connectome has two fundamental components: the nodes and the connections between them. While much attention has been given to deriving atlases and measuring the connections between nodes, there have been no studies examining the networks within nodes. Here we demonstrate that each node contains significant connectivity information, that varies systematically across task-induced states and subjects, such that measures based on these variations can be used to classify tasks and identify subjects. The results are not specific for any particular atlas but hold across different atlas resolutions. To date, studies examining changes in connectivity have focused on edge changes and assumed there is no useful information within nodes. Our findings illustrate that for typical atlases, within-node changes can be significant and may account for a substantial fraction of the variance currently attributed to edge changes.

Overset Grid Assembler and Flow Solver with Adaptive Spatial Load Balancing

An overset mesh approach is useful for unsteady flow problems which involve components moving relative to each other. Since the generation of a single mesh around all components is prone to mesh stretching due to the relative motion of bodies, using the overset grid methodology, an individual mesh can be generated for each component. In this study, a parallel overset grid assembler was developed to establish connectivity across component meshes. Connectivity information was transferred to the developed parallel flow solver. The assembler uses multiple methods such as alternating digital tree and stencil walking to reduce the time spent on domain connectivity. Both the assembler and solver were partitioned spatially so that overlapping mesh blocks reside in the same partitions. Spatial partitioning was performed using a 3D space partitioning structure, namely octree, to which mesh blocks are registered. The octree was refined adaptively until bins of octree could be evenly distributed to processors. The assembler and solver were tested on a generic helicopter configuration in terms of load balance, scalability, and memory usage.

A Theoretical Analysis of Mobility Detection in Connectivity-Based Localization for Short-Range Networks

This paper presents a theoretical analysis of mobility detection in connectivity-based localization, which exploits connectivity information as range measurements to anchors at a known location, to investigate how well and how precise mobility can be detected with connectivity in short-range networks. We derive mobility detection, miss detection, and false alarm probabilities in terms of a mobility detection threshold, defined as the minimum distance to detect the mobility, under the shadow fading channel and arbitrary mobility models to take into account practical and general scenarios. Based on the derivations, we address the threshold determination with the criteria in the sense of the minimum average error from miss detection and false alarm. Numerical and simulation evaluations are performed to verify our theoretical derivations, to show that increasing anchor numbers can improve the mobility detection probability with a smaller detection threshold, and that the probabilities are bounded by the weights of miss detection and false alarm. This work is the first attempt at addressing the performance of mobility detection using connectivity, and it can be utilized as a baseline for connectivity-based mobility tracking.

Unsupervised Structural Graph Node Representation Learning

Unsupervised Graph Representation Learning methods learn a numerical representation of the nodes in a graph. The generated representations encode meaningful information about the nodes' properties, making them a powerful tool for tasks in many areas of study, such as social sciences, biology or communication networks. These methods are particularly interesting because they facilitate the direct use of standard Machine Learning models on graphs. Graph representation learning methods can be divided into two main categories depending on the information they encode, methods preserving the nodes connectivity information, and methods preserving nodes' structural information. Connectivity-based methods focus on encoding relationships between nodes, with neighboring nodes being closer together in the resulting latent space. On the other hand, structure-based methods generate a latent space where nodes serving a similar structural function in the network are encoded close to each other, independently of them being connected or even close to each other in the graph. While there are a lot of works that focus on preserving nodes' connectivity information, only a few works study the problem of encoding nodes' structure, specially in an unsupervised way. In this dissertation, we demonstrate that properly encoding nodes' structural information is fundamental for many real-world applications, as it can be leveraged to successfully solve many tasks where connectivity-based methods fail. One concrete example is presented first. In this example, the task consists of detecting malicious entities in a real-world financial network. We show that to solve this problem, connectivity information is not enough and show how leveraging structural information provides considerable performance improvements. This particular example pinpoints the need for further research on the area of structural graph representation learning, together with the limitations of the previous state-of-the-art. We use the acquired knowledge as a starting point and inspiration for the research and development of three independent unsupervised structural graph representation learning methods: Structural Iterative Representation learning approach for Graph Nodes (SIR-GN), Structural Iterative Lexicographic Autoencoded Node Representation (SILA), and Sparse Structural Node Representation (SparseStruct). We show how each of our methods tackles specific limitations on the previous state-of-the-art on structural graph representation learning such as scalability, representation meaning, and lack of formal proof that guarantees the preservation of structural properties. We provide an extensive experimental section where we compare our three proposed methods to the current state-of-the-art on both connectivity-based and structure-based representation learning methods. Finally, in this dissertation, we look at extensions of the basic structural graph representation learning problem. We study the problem of temporal structural graph representation. We also provide a method for representation explainability.

Understanding signaling and metabolic paths using semantified and harmonized information about biological interactions

BackgroundTo grasp the complexity of biological processes, the biological knowledge is often translated into schematic diagrams of biological pathways, such as signalling and metabolic pathways. These pathway diagrams describe relevant connections between biological entities and incorporate domain knowledge in a visual format that is easier for humans to interpret. It has already been established that these diagrams can be represented in machine readable formats, as done in KEGG, Reactome, and WikiPathways. However, while humans are good at interpreting the message of the creator of such a diagram, algorithms struggle when the diversity in drawing approaches increases. WikiPathways supports multiple drawing styles, and therefore needs to harmonize this to offer semantically enriched access via the Resource Description Framework format. Particularly challenging in the normalization of diagrams are the interactions between the biological entities, so that we can glean information about the connectivity of the entities represented. These interactions include information about the type of interaction (metabolic conversion, inhibition, etc.), the direction, and the participants. Availability of the interactions in a semantic and harmonized format enables searching the full network of biological interactions and integration with the linked data cloud.ResultsWe here study how the graphically modelled biological knowledge in diagrams can be semantified and harmonized efficiently, and exemplify how the resulting data can be used to programmatically answer biological questions. We find that we can translate graphically modelled biological knowledge to a sufficient degree into a semantic model of biological knowledge and discuss some of the current limitations. Furthermore, we show how this interaction knowledge base can be used to answer specific biological questions.ConclusionThis paper demonstrates that most of the graphical biological knowledge from WikiPathways is modelled in the semantic layer of WikiPathways with the semantic information intact and connectivity information preserved. The usability of the WikiPathways pathway and connectivity information has shown to be useful and has been integrated into other platforms. Being able to evaluate how biological elements affect each other is useful and allows, for example, the identification of up or downstream targets that will have a similar effect when modified.

MNCE: Multi-Hop Node Localization Algorithm for Wireless Sensor Network Based on Error Correction

Considering the problems of large error and high localization costs of current range-free localization algorithms, a MNCE algorithm based on error correction is proposed in this study. This algorithm decomposes the multi-hop distance between nodes into several small hops. The distance of each small hop is estimated by using the connectivity information of adjacent nodes; small hops are accumulated to obtain the initial estimated distance. Then, the error-correction rate based on the error-correction concept is proposed to correct the initial estimated distance. Finally, the location of the target node is resolved by total least square methods, according to the information on the anchor nodes and estimated distances. Simulation experiments show that the MNCE algorithm is superior to the similar types of localization algorithms.