Search

Federated learning (FL), with the aim of training machine learning models using data and computational resources on edge devices without sharing raw local data, is essential for improving agricultural management and smart agriculture. This study is a review of FL applications that address various agricultural problems. We compare the types of data partitioning and types of FL (horizontal partitioning and horizontal FL, vertical partitioning and vertical FL, and hybrid partitioning and transfer FL), architectures (centralized and decentralized), levels of federation (cross-device and cross-silo), and the use of aggregation algorithms in different reviewed approaches and applications of FL in agriculture. We also briefly review how the communication challenge is solved by different approaches. This work is useful for gaining an overview of the FL techniques used in agriculture and the progress made in this field. [ABSTRACT FROM AUTHOR]

One of the most important problems in complex networks is the location of nodes that are essential or play a main role in the network. Nodes with main local roles are the centers of real communities. Communities are sets of nodes of complex networks and are densely connected internally. Choosing the right nodes as seeds of the communities is crucial in determining real communities. We propose a new centrality measure named density-based entropy centrality for the local identification of the most important nodes. It measures the entropy of the sum of the sizes of the maximal cliques to which each node and its neighbor nodes belong. The proposed centrality is a local measure for explaining the local influence of each node, which provides an efficient way to locally identify the most important nodes and for community detection because communities are local structures. It can be computed independently for individual vertices, for large networks, and for not well-specified networks. The use of the proposed density-based entropy centrality for community seed selection and community detection outperforms other centrality measures. [ABSTRACT FROM AUTHOR]

The paper reviews community detection methods used besides other partition algorithms for power grid partitioning. The power grid is a complex network. A complex network forms a community structure. There are many edges between community nodes and relatively few of those connecting the vertices of different communities. Community detection is an optimization problem. Its goal is to detect communities that belong to a network or graph in which nodes of a particular community in the same community share properties of the same characteristics or functional relationships. Power grids use adaptable community detection algorithms of various suitability rates used in power grid partitioning. The paper presents types and adaptations of various community detection algorithms used in power grid partitioning. [ABSTRACT FROM AUTHOR]

• Interpolative coding is efficient at chain code compression. • Interpolative coding is improved with Golomb coding. • Proper selection of Golomb parameter yields slightly better compression. A lossless chain code compression algorithm consisting of Burrows-Wheeler Transform, Move-To-Front Transform, and a modified interpolative coding is presented in this paper. The interpolative coding divides a strictly increasing array of integers and encodes the middle element recursively, where the required number of bits is obtained from the difference of the array's border values. In this paper, it is shown that, for some intervals, Golomb coding is more efficient than the interpolative coding. This fact was used to construct a hybrid approach that yields slightly better compression results than the state-of-the-art while compressing Three OrThogonal chain code, Freeman chain code in four and eight directions, and Vertex Chain Code. [ABSTRACT FROM AUTHOR]

Automatic network clustering is an important method for mining the meaningful communities of complex networks. Uncovered communities help to understand the potential system structure and functionality. Many algorithms that use multiple optimization criteria and optimize a population of solutions are difficult to apply to real systems because they suffer a long optimization process. In this paper, in order to accelerate the optimization process and to uncover multiple significant community structures more effectively, a multi-objective evolutionary algorithm is proposed and evaluated using problem-specific genetic mutation and group crossover, and problem-specific initialization. Since crossover operators mainly contribute to performance of genetic algorithms, more problem-specific group crossover operators are introduced and evaluated for intelligent evolution of population. The experiments on both artificial and real-world networks demonstrate that the proposed evolutionary algorithm with problem-specific genetic operations has effective performance on discovering the community structure of networks. [ABSTRACT FROM AUTHOR]

Community detection in large networks has been a focus of recent research in many of fields, including biology, physics, social sciences, and computer science. Most community detection methods partition the entire network into communities, groups of nodes that have many connections within communities and few connections between them and do not identify different roles that nodes can have in communities. We propose a community detection model that integrates more different measures that can fast identify communities of different sizes and densities. We use node degree centrality, strong similarity with one node from community, maximal similarity of node to community, compactness of communities and separation between communities. Each measure has its own strength and weakness. Thus, combining different measures can benefit from the strengths of each one and eliminate encountered problems of using an individual measure. We present a fast local expansion algorithm for uncovering communities of different sizes and densities and reveals rich information on input networks. Experimental results show that the proposed algorithm is better or as effective as the other community detection algorithms for both real-world and synthetic networks while it requires less time. [ABSTRACT FROM AUTHOR]

Abstract: Cluster validity indices are used to validate results of clustering and to find a set of clusters that best fits natural partitions for given data set. Most of the previous validity indices have been considerably dependent on the number of data objects in clusters, on cluster centroids and on average values. They have a tendency to ignore small clusters and clusters with low density. Two cluster validity indices are proposed for efficient validation of partitions containing clusters that widely differ in sizes and densities. The first proposed index exploits a compactness measure and a separation measure, and the second index is based an overlap measure and a separation measure. The compactness and the overlap measures are calculated from few data objects of a cluster while the separation measure uses all data objects. The compactness measure is calculated only from data objects of a cluster that are far enough away from the cluster centroids, while the overlap measure is calculated from data objects that are enough near to one or more other clusters. A good partition is expected to have low degree of overlap and a larger separation distance and compactness. The maximum value of the ratio of compactness to separation and the minimum value of the ratio of overlap to separation indicate the optimal partition. Testing of both proposed indices on some artificial and three well-known real data sets showed the effectiveness and reliability of the proposed indices. [ABSTRACT FROM AUTHOR]

Abstract: This paper presents an agglomerative hierarchical clustering algorithm for spatial data. It discovers clusters of arbitrary shapes which may be nested. The algorithm uses a sweeping approach consisting of three phases: sorting is done during the preprocessing phase, determination of clusters is performed during the sweeping phase, and clusters are adjusted during the post processing phase. The properties of the algorithm are demonstrated by examples. The algorithm is also adapted to the streaming algorithm for clustering large spatial datasets. [Copyright &y& Elsevier]

Abstract: This paper introduces k′-means algorithm that performs correct clustering without pre-assigning the exact number of clusters. This is achieved by minimizing a suggested cost-function. The cost-function extends the mean-square-error cost-function of k-means. The algorithm consists of two separate steps. The first is a pre-processing procedure that performs initial clustering and assigns at least one seed point to each cluster. During the second step, the seed-points are adjusted to minimize the cost-function. The algorithm automatically penalizes any possible winning chances for all rival seed-points in subsequent iterations. When the cost-function reaches a global minimum, the correct number of clusters is determined and the remaining seed points are located near the centres of actual clusters. The simulated experiments described in this paper confirm good performance of the proposed algorithm. [Copyright &y& Elsevier]

This paper introduces a novel approach to increase the spatiotemporal resolution of an arbitrary environmental variable. This is achieved by utilizing machine learning algorithms to construct a satellite-like image at any given time moment, based on the measurements from IoT sensors. The target variables are calculated by an ensemble of regression models. The observed area is gridded, and partitioned into Voronoi cells based on the IoT sensors, whose measurements are available at the considered time. The pixels in each cell have a separate regression model, and take into account the measurements of the central and neighboring IoT sensors. The proposed approach was used to assess NO 2 data, which were obtained from the Sentinel-5 Precursor satellite and IoT ground sensors. The approach was tested with three different machine learning algorithms: 1-nearest neighbor, linear regression and a feed-forward neural network. The highest accuracy yield was from the prediction models built with the feed-forward neural network, with an R M S E of 15.49 × 10 − 6 mol/m 2 . [ABSTRACT FROM AUTHOR]

This paper proposes a new string transformation technique called Move with Interleaving (MwI). Four possible ways of rearranging 2D raster images into 1D sequences of values are applied, including scan-line, left-right, strip-based, and Hilbert arrangements. Experiments on 32 benchmark greyscale raster images of various resolutions demonstrated that the proposed transformation reduces information entropy to a similar extent as the combination of the Burrows–Wheeler transform followed by the Move-To-Front or the Inversion Frequencies. The proposed transformation MwI yields the best result among all the considered transformations when the Hilbert arrangement is applied. [ABSTRACT FROM AUTHOR]

This paper introduces a new algorithm for Boolean operations on rasterized geometric shapes that are represented with chain codes. The algorithm works in three steps. Firstly, the chain code symbols are transformed in the Hilbert space, where the overlaid chain code symbols are recognised. After that, a suitable starting cell is determined. Finally, the walk-about through the sequence of the initial chain code symbols is performed to obtain the sequence of chain code symbols representing the shape of the required Boolean operation. The algorithm is demonstrated on Freeman chain code in four directions. The time and space complexity of the proposed algorithm is linear, which was proven theoretically and confirmed by experiments. [ABSTRACT FROM AUTHOR]

Clustering is one of the better known unsupervised learning methods with the aim of discovering structures in the data. This paper presents a distance-based Sweep-Hyperplane Clustering Algorithm (SHCA), which uses sweep-hyperplanes to quickly locate each point's approximate nearest neighbourhood. Furthermore, a new distance-based dynamic model that is based on 2N-tree hierarchical space partitioning, extends SHCA's capability for finding clusters that are not well-separated, with arbitrary shape and density. Experimental results on different synthetic and real multidimensional datasets that are large and noisy demonstrate the effectiveness of the proposed algorithm. [ABSTRACT FROM AUTHOR]

A new method for encoding a sequence of integers, named Binary Adaptive Sequential Coding with Return to Bias , is proposed in this paper. It extends the compressing pipeline for chain codes' compression consisting of Burrows Wheeler Transform, Move-To-Front Transform, and Adaptive Arithmetic Coding. We also explain when to include the Zero-Run Transform into the above-mentioned pipeline. The Zero-Run Transform generates a sequence of integers corresponding to the number of zero-runs. This sequence is encoded by Golomb coding, Binary Adaptive Sequential Coding, and the new Binary Adaptive Sequential Coding with Return to Bias. Finally, a comparison is performed with the two state-of-the-art methods. The proposed method achieved similar compression efficiency for the Freeman chain code in eight directions. However, for the chain codes with shorter alphabets (Freeman chain code in four directions, Vertex Chain Code, and Three-OrThogonal chain code), the introduced method outperforms the referenced ones. [ABSTRACT FROM AUTHOR]