Appropriate Number Of Classes Research Articles

Temporal Development GAN (TD-GAN): Crafting More Accurate Image Sequences of Biological Development

In this study, we propose a novel Temporal Development Generative Adversarial Network (TD-GAN) for the generation and analysis of videos, with a particular focus on biological and medical applications. Inspired by Progressive Growing GAN (PG-GAN) and Temporal GAN (T-GAN), our approach employs multiple discriminators to analyze generated videos at different resolutions and approaches. A new Temporal Discriminator (TD) that evaluates the developmental coherence of video content is introduced, ensuring that the generated image sequences follow a realistic order of stages. The proposed TD-GAN is evaluated on three datasets: Mold, Yeast, and Embryo, each with unique characteristics. Multiple evaluation metrics are used to comprehensively assess the generated videos, including the Fréchet Inception Distance (FID), Frechet Video Distance (FVD), class accuracy, order accuracy, and Mean Squared Error (MSE). Results indicate that TD-GAN significantly improves FVD scores, demonstrating its effectiveness in generating more coherent videos. It achieves competitive FID scores, particularly when selecting the appropriate number of classes for each dataset and resolution. Additionally, TD-GAN enhances class accuracy, order accuracy, and reduces MSE compared to the default model, demonstrating its ability to generate more realistic and coherent video sequences. Furthermore, our analysis of stage distribution in the generated videos shows that TD-GAN produces videos that closely match the real datasets, offering promising potential for generating and analyzing videos in different domains, including biology and medicine.

A Self-Supervised Learning Approach for Accelerating Wireless Network Optimization

The prevailing issue in multi-hop wireless networking is interference management, which militates against the efficiency of traditional routing and scheduling algorithms. We develop a self-supervised learning approach to address the classic NP-hard problem of capacity optimization over a multi-hop wireless network, where the routing and scheduling decisions are deeply coupled. Our two-stage design leverages historical computation experiences to accelerate the optimization of new problem instances, where an instance represents the application-level input containing network topology, interference model, and other user-level traffic constraints. The first stage, Scheduling Structure Classification (SSC), distills the scheduling structure of the historical optimization instances into an appropriate number of classes, through a properly designed clustering algorithm without prior assumption or knowledge. The second stage uses the instances labelled with class information from the first stage to train an Application Identification (AID) neural network model capable of predicting a future problem instance's scheduling class given its application-level information. When solving the new instance, its predicted scheduling class and the associated scheduling structure are exploited to compute an efficient approximate solution, avoiding the time-consuming iterative search for such scheduling structure as practiced by the conventional approaches. We apply our method to different types of wireless multi-commodity flow problems across various network sizes and disparate flow requirements. The results demonstrate that our method significantly reduces the computation time robustly by at least 70% with only slight loss in the solution quality.

Social inequalities in multimorbidity patterns in southern Spain: findings from the DEMMOCAD survey

Abstract Background Multimorbidity (MM) is associated with lower quality of life, greater disability, and higher use of health services. It is a complex problem that is difficult to capture due to the broad spectrum of concurrent chronic diseases involved. There is a need to identify and characterize patterns of chronic conditions in the local context of specific population groups. The DEMMOCAD project aims to respond to this knowledge gap by detecting patterns of MM and their inequalities in the province of Cadiz (Spain). Methods A cross-sectional study was carried out by means of telephone interviews with people over 50 years of age. The final sample was 1592 individuals with MM. A latent class analysis was carried out to identify patterns of MM from 31 chronic conditions. First, the appropriate number of classes was established, considering model fit indices, class size, and clinical interpretability. Subsequently, covariates were introduced into the model using the three-step approach, a technique that minimizes biases in the multinomial regression model. Results Preliminary analyses of the goodness-of-fit indices of the model derived five MM patterns (entropy = 0.727): (C1) mild MM; (C2) cardiovascular; (C3) musculoskeletal; (C4) musculoskeletal plus mental; and (C5) complex MM. Compared with class C1, persons in class C5 were significantly older and less educated, class C4 had a lower income, class C3 was smokers and disabled, and class C2 was characteristic among older males. All four classes also showed lower scores on mental and physical dimensions of the SF12 scale compared to class C1. Conclusions In addition to providing an adjusted characterization of the population of the area analyzed, these initial findings highlight the existence of social inequalities in multimorbidity at the local level that should be addressed by implementing policies targeting the most vulnerable groups in Cadiz (low socioeconomic status groups, people with disabilities, and the elderly). Key messages

Partition of GB-InSAR deformation map based on dynamic time warping and $k$-means

Ground-based interferometric synthetic aperture radar (GB-InSAR) can take deformation measurement with a high accuracy. Partition of the GB-InSAR deformation map benefits analyzing the deformation state of the monitoring scene better. Existing partition methods rely on labelled datasets or single deformation feature, and they cannot be effectively utilized in GB-InSAR applications. This paper proposes an improved partition method of the GB-InSAR deformation map based on dynamic time warping (DTW) and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$k$</tex> -means. The DTW similarities between a reference point and all the measurement points are calculated based on their time-series deformations. Then the DTW similarity and cumulative deformation are taken as two partition features. With the <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$k$</tex> -means algorithm and the score based on multi evaluation indexes, a deformation map can be partitioned into an appropriate number of classes. Experimental datasets of West Copper Mine are processed to validate the effectiveness of the proposed method, whose measurement points are divided into seven classes with a score of 0.315 1.

Effects of Kindergarten, Family Environment, and Physical Activity on Children's Physical Fitness.

To explore the relationship between kindergarten environmental factors, children's physical activity, and physical fitness, this study uses the stratified random sampling method to obtain 4,600 children in relevant kindergartens. The questionnaire survey and children's physical fitness test were completed with the help of parents and kindergarten staff. The exploratory (EFA) and confirmatory (CFA) factor analysis is used to process the obtained database and set the significance level of all indicators α = 0.05. The results show that kindergarten environmental factors significantly affect children's physical activity and healthy physical fitness. Children with large play areas in these kindergartens, more sports equipment items, who participate in more than three games per week, of no <40 min of each class, with an appropriate number of classes, and excellent teachers' teaching ability have better physical fitness. Family environmental factors significantly affect children's physical activity and fitness. Children with more family sports equipment items, more peers living nearby, safer playing places, more hands and feet, and parents who are good at sports have better performance in health fitness. Children's physical activity not only directly affects their performance of physical fitness, but also plays a dual intermediary role between kindergarten environment and physical fitness, family environment, and healthy physical fitness.

Beyond the Toxic Trio: Exploring Demand Typologies in Children’s Social Care

AbstractDemand for children’s social care is often conflated with rates of intervention and associated with a limited constellation of parental risk factors. This article reports on a more comprehensive picture of demand obtained through a quantitative study of child welfare interventions in England. Longitudinal child-level data were combined from children’s social care services in six English local authorities over a four-year period (2015–2018). Latent class analysis was undertaken for a random sample of child episodes where an assessment was undertaken (n = 15,000). The results were tested for consistency across LAs and to identify the most appropriate number of classes. Conditional probabilities were used to interpret the demand represented by each class, and to explore the relationship between typologies and child characteristics such as age, gender and ethnicity. The analysis found seven classes, or typologies of demand, to be present in factors at assessment across all the LAs, which were linked to certain child characteristics and intervention pathways. The findings go beyond the ‘toxic trio’ terminology often used to profile risks to children and support the innovative use of administrative data to provide insight into patterns of demand. Implications are discussed for strategic responses to child welfare problems and the multi-agency context of prevention.

FORMAL REPRESENTATION OF THE PIXEL-BY-PIXEL CLASSIFICATION PROCESS USING A MODIFIED WANG-MENDEL NEURAL NETWORK

The subject of research in the article are the processes of formalization of the pixel-by-pixel classification problem using the modified fuzzy neural production network of Wang-Mendel for segmentation of urban structures in the automated analysis of space and aerial photographs of the city. The purpose of the work is to develop the architecture of the modified fuzzy neural production network of Wang-Mendel as a classifier for image segmentation to increase the values of efficiency and reliability of urban monitoring. The following tasks are solved in the article: analysis of possibilities of Wang-Mendel network modification based on representation of membership functions in terms of interval fuzzy sets of the second type (IFST2) and realization of phasing, aggregation and activation operations using IFST 2 operations, development of the architecture of the modified fuzzy neural production network of Wang-Mendel as a classifier for image segmentation. The following methods and models are used: methods and models of fuzzy set theory (fuzzy Wang-Mendel neural network, interval fuzzy sets of the second type), methods and models of deep learning methodology (convolutional neural network for image segmentation (auto coder) U-net). The following results were obtained: the use of a fuzzy Wang-Mendel neural network as a classifier of a modified U-Net decoder based on the representation of membership functions in IFST2 and the implementation of phasing, aggregation and activation operations using operations on IFST2; introduction of an additional operation of type reduction in the phase of dephasification of the original variable based on the classical method of the center of gravity (centroid); introduction of several outputs of the network to recognize the appropriate number of classes (subclasses) of the subject area. To do this, the third layer is represented as a set of several pairs of adder neurons, and the fourth implements several normalizing neurons, the number of which corresponds to the number of pairs of the third layer. Conclusions: the use in the architecture of a convolutional neural network for segmentation of U-net images as a classifier of the modified fuzzy neural production network of Wang-Mendel will provide an additional increase in the accuracy of pixel-by-pixel classification of certain objects. Instead of fuzzy sets of the first type (FST1) in this network IFST2 are used. The proposed IFST2, on the one hand, provide a formalization of more additional degrees of uncertainty compared to FST1, on the other hand, are "implemented" in the development of fuzzy systems (models) and have less computational complexity, compared to fuzzy sets of the second type (FST2).

Improved Frequency Table with Application to Environmental Data

This paper proposes three different statistics to be used to represent the magnitude observations in each class when estimating the statistical measures from the frequency table for continuous data. The existing frequency tables use the midpoint as the magnitude of observations in each class, which results in an error called grouping error. Using the midpoint is due to the assumption that the observations in each class are uniformly distributed and concentrated around their midpoint, which is not always valid. In this research, construction of the frequency tables using the three proposed statistics, the arithmetic mean, median, and midrange and midpoint are respectively named, Method 1, Method 2, Method 3, and the Existing method. The four methods are compared using root-mean-squared error (RMSE) by performing simulation studies using three distributions, normal, uniform, exponential distributions. The simulation results are validated using real data, Glasgow weather data. The findings indicated that using the arithmetic mean to represent the magnitude of observations in each class of the frequency table leads to minimal error relative to other statistics. It is followed by using the median, for data simulated from normal and exponential distributions, and using midrange for data simulated from uniform distribution. Meanwhile, in choosing the appropriate number of classes used in constructing the frequency tables, among seven different rules used, the freedman and Diaconis rule is the recommended rule.

Texture Features Extraction and Backtracking Search Optimization Algorithm for Satellite Image Clustering

Clustering is one of the data analysis methods used in image processing. It consists in dividing all the pixels of an image into a corresponding class pixel by pixel. In this paper, the Backtracking Search Algorithm (BSA) has been investigated in order to cluster this heterogeneous set of information using a combination of different textural features. The validity index is used to determine the appropriate number of classes representing an image and to evaluate the results of the classification algorithm. The study realized with Calinski Harabasz (CHI) and Davies-Bouldin Index (DBI) allows finding the best validity index to evaluate fitness function. The experiences obtained show the effectiveness and performance of the Davies-Bouldin Index (DBI) with BSA algorithm.

Deciding on the Starting Number of Classes of a Latent Class Tree.

In recent studies, latent class tree (LCT) modeling has been proposed as a convenient alternative to standard latent class (LC) analysis. Instead of using an estimation method in which all classes are formed simultaneously given the specified number of classes, in LCT analysis a hierarchical structure of mutually linked classes is obtained by sequentially splitting classes into two subclasses. The resulting tree structure gives a clear insight into how the classes are formed and how solutions with different numbers of classes are substantively linked to one another. A limitation of the current LCT modeling approach is that it allows only for binary splits, which in certain situations may be too restrictive. Especially at the root node of the tree, where an initial set of classes is created based on the most dominant associations present in the data, it may make sense to use a model with more than two classes. In this article, we propose a modification of the LCT approach that allows for a nonbinary split at the root node, and we provide methods to determine the appropriate number of classes in this first split, based either on theoretical grounds or on a relative improvement of fit measure. This novel approach also can be seen as a hybrid of a standard LC model and a binary LCT model, in which an initial, oversimplified but interpretable model is refined using an LCT approach. Furthermore, we show how to apply an LCT model when a nonstandard LC model is required. These new approaches are illustrated using two empirical applications: one on social capital and the other on (post)materialism.

Semiautomated Segmentation of Sentinel-1 SAR Imagery for Mapping Sea Ice in Labrador Coast

This study aims at proposing a semiautomated sea ice segmentation workflow utilizing Sentinel-1 synthetic aperture radar imagery. The workflow consists of two main steps. First, preferable features in sea ice interpretation were determined with a random forest feature selection method. Second, an unsupervised graph-cut image segmentation was performed. The workflow was tested on 13 Sentinel-1A images from January to June 2016, and the results were evaluated on open water segmentation per ice charts provided by Canada Ice Service. The results showed that the proposed workflow was able to segment Sentinel-1 images in to appropriate number of classes, and the potential water identification rate reached 95%.

How to reverse first time donors to become regular donors? A questionnaire survey in Estonia

BackgroundDonating blood in Estonia is non-remunerated and voluntary. Estonian Blood Service system has four independent regional blood centres that are responsible for blood collection, processing, screening and distribution of blood components to hospitals for clinical use. Study designThis research was carried out as a questionnaire survey. A questionnaire was developed to study lapsing first time donors’ (FTD) blood donation experience, intention and willingness to donate again. MethodsA thousand five hundred and forty-six questionnaires were posted to donors who had one successful donation in 2010 and who had not returned to second donation till the year 2012. For data analysis routine statistical methods were used. To evaluate the most appropriate number of classes, based on previous experience and future expectations, latent class analysis was used. ResultsThere were 453 respondents (29.3%). For the majority of aspects of blood donation experience the emotions were positive. Results of the study suggested that blood collection agencies should intervene to bolster donors’ attitudes, perceived control, and identity as a donor during this crucial post-first donation period. ConclusionFirst blood donation seems to have been a positive experience. Reasons leading to stopping blood donation should be studied further. Establishing a donor registry for Estonia would be essential to keep track of donors.

Comparison of the Accuracy and Performance of Different Numbers of Classes in Discretised Solution Method for Population Balance Model

One way of solving population balance model (PBM) in a time efficient way is by means of discretisation of the population property of interest. A computational grid, for example, vi+1=kvi (vi is the volume of particle in class i), could be used to classify the particles in discretisation techniques. However, there are still disagreements in the appropriate number of classes divided by the grids. In this study, the different numbers of classes for solving PBM were compared in terms of accuracy and performance to describe the particle size distribution (PSD) from the flocculation of activated sludge. It is found that the simulated PSDs are similar to the experimental data for all the geometric grids (vi+1:vi≤2), and there is no obvious difference among the values of calibrated parameter, ratio of breakage rate coefficient and collision efficiency, for each velocity gradient. However, the simulation results with less error could be obtained with larger number of classes, and more computational times, which show exponential relationship with the number of classes, are needed. Considering numerical accuracy and efficiency, the classes 35 or a geometric grid with factor 1.6, aligning with the Fibonacci sequence (vi+vi-1≈vi+1), is recommended for the particles in the size range of 5.5~1086 μm.

Hydrologic landscape regionalisation using deductive classification and random forests.

Landscape classification and hydrological regionalisation studies are being increasingly used in ecohydrology to aid in the management and research of aquatic resources. We present a methodology for classifying hydrologic landscapes based on spatial environmental variables by employing non-parametric statistics and hybrid image classification. Our approach differed from previous classifications which have required the use of an a priori spatial unit (e.g. a catchment) which necessarily results in the loss of variability that is known to exist within those units. The use of a simple statistical approach to identify an appropriate number of classes eliminated the need for large amounts of post-hoc testing with different number of groups, or the selection and justification of an arbitrary number. Using statistical clustering, we identified 23 distinct groups within our training dataset. The use of a hybrid classification employing random forests extended this statistical clustering to an area of approximately 228,000 km2 of south-eastern Australia without the need to rely on catchments, landscape units or stream sections. This extension resulted in a highly accurate regionalisation at both 30-m and 2.5-km resolution, and a less-accurate 10-km classification that would be more appropriate for use at a continental scale. A smaller case study, of an area covering 27,000 km2, demonstrated that the method preserved the intra- and inter-catchment variability that is known to exist in local hydrology, based on previous research. Preliminary analysis linking the regionalisation to streamflow indices is promising suggesting that the method could be used to predict streamflow behaviour in ungauged catchments. Our work therefore simplifies current classification frameworks that are becoming more popular in ecohydrology, while better retaining small-scale variability in hydrology, thus enabling future attempts to explain and visualise broad-scale hydrologic trends at the scale of catchments and continents.

A clustering-based model for class responsibility assignment problem in object-oriented analysis

Assigning responsibilities to classes is a vital task in object-oriented analysis and design, and it directly affects the maintainability and reusability of software systems. There are many methodologies to help recognize the responsibilities of a system and assign them to classes, but all of them depend greatly on human judgment and decision-making. In this paper, we propose a clustering-based model to solve the class responsibility assignment (CRA) problem. The proposed model employs a novel interactive graph-based method to find inheritance hierarchies, and two novel criteria to determine the appropriate number of classes. It reduces the dependency of CRA on human judgment and provides a decision-making support for CRA in class diagrams. To evaluate the proposed model, we apply three different hierarchical agglomerative clustering algorithms and two different types of similarity measures. By comparing the obtained results of clustering techniques with the models designed by multi-objective genetic algorithm (MOGA), it is revealed that clustering techniques yield promising results.

Unsupervised Clustering of Images Using Harmony Search Algorithm

Clustering plays an important role in the image processing. It permits to assign a label to each point of the image from a collection of defined classes. Among the domains that use the clustering, we can mention the Remote Sensing for identification of different regions constituting a satellite image. Evaluation of the clustering algorithm results is based on the validity index. In this paper, we applied the Harmony Search algorithm (HS) for make an unsupervised clustering. Thereafter, we evaluated the performance of this tool by analyzing the results obtained. These results show that the validity index determines automatically the appropriate number of classes that represent an image. The study realized with several validity indices allowed us to find the best validity index to evaluate the performance and robustness of the algorithm HS. The experiences obtained with this algorithm show the effectiveness and performance in the stable clustering for given problem.

An Unsupervised Classification Approach for Polarimetric SAR Data Based on the Chernoff Distance for Complex Wishart Distribution

A new unsupervised classification approach for polarimetric synthetic aperture radar (POLSAR) data is proposed in this paper. The Wishart-Chernoff distance is calculated and used in an agglomerative hierarchical clustering approach. Initial segmentation of POLSAR data into clusters is obtained based on the total backscattering power (SPAN) combined with the entropy, alpha angle, and anisotropy. The complex Wishart clustering is performed to optimize the initialization. Optimized clusters with minimum Wishart-Chernoff distance are merged hierarchically into an appropriate number of classes. The appropriate number of classes is estimated based on the data log-likelihood algorithm. Classification results show that the use of Wishart-Chernoff distance is superior to that of the Wishart test statistic distance. The effectiveness of the proposed Wishart-Chernoff distance is demonstrated using Advanced Land Observing Satellite POLSAR data.

Comparing matrix distance measures for unsupervised POLSAR data classification of sea ice based on agglomerative clustering

Clustering is a technique that can be applied for unsupervised classification of polarimetric synthetic aperture radar (POLSAR) data, an important analysis technique of radar polarimetry. Six matrix distance measures have been investigated and compared through an agglomerative clustering of RADARSAT-2 POLSAR data. The considered matrix distance measures were used as similarity criteria for merging clusters hierarchically into an appropriate number of classes. In this study, the considered distances are Manhattan, Euclidean, Bartlett, revised Wishart, Wishart test statistic, and Wishart Chernoff. Results show that the Bartlett, revised Wishart, and Wishart Chernoff distances all produce identical classification results. The Manhattan distance retrieved classification results close to those obtained by the Euclidean distance. The Bartlett, revised Wishart, and Wishart Chernoff distances produced the most accurate classification results. The study area is located in Hudson Bay offshore Churchill, Manitoba, Canada.

Bayesian inference and model selection in latent class logit models with parameter constraints: An application to market segmentation

Latent class models have recently drawn considerable attention among many researchers and practitioners as a class of useful tools for capturing heterogeneity across different segments in a target market or population. In this paper, we consider a latent class logit model with parameter constraints and deal with two important issues in the latent class models--parameter estimation and selection of an appropriate number of classes--within a Bayesian framework. A simple Gibbs sampling algorithm is proposed for sample generation from the posterior distribution of unknown parameters. Using the Gibbs output, we propose a method for determining an appropriate number of the latent classes. A real-world marketing example as an application for market segmentation is provided to illustrate the proposed method.

Coastal typology development with heterogeneous data sets

This paper presents a data-driven expert-guided method of coastal typology development using a large, heterogeneous data set. The development of coastal typologies is driven by a desire to upscale detailed regional information to a global scale in order to study coastal zone function and the effects of global climate change. We demonstrate two methods of automatic typology generation – unsupervised clustering and region growing with agglomerative clustering – and a method of selecting an appropriate number of classes based on the concept of Minimum Description Length. We compare two methods of defining distance between data points with a large number of variables and potentially missing data – average scaled Euclidean distance and maximum scaled difference. To visualize the resulting typologies we use a novel algorithm for assigning colors to different classes of data based on class similarity in a high-dimensional space. This combination of techniques results in a methodology through which one or more experts can easily develop a useful coastline typology with results that are similar to preexisting expert typologies, but which makes the process more quantitative, objective, consistent, and applicable across space and time.