Probability-based Methods Research Articles

Identifying Potential Natural Antibiotics from Unani Formulas through Machine Learning Approaches.

The Unani Tibb is a medical system of Greek descent that has undergone substantial dissemination since the 11th century and is currently prevalent in modern South and Central Asia, particularly in primary health care. The ingredients of Unani herbal medicines are primarily derived from plants. Our research aimed to address the pressing issues of antibiotic resistance, multi-drug resistance, and the emergence of superbugs by examining the molecular-level effects of Unani ingredients as potential new natural antibiotic candidates. We utilized a machine learning approach to tackle these challenges, employing decision trees, kernels, neural networks, and probability-based methods. We used 12 machine learning algorithms and several techniques for preprocessing data, such as Synthetic Minority Over-sampling Technique (SMOTE), Feature Selection, and Principal Component Analysis (PCA). To ensure that our model was optimal, we conducted grid-search tuning to tune all the hyperparameters of the machine learning models. The application of Multi-Layer Perceptron (MLP) with SMOTE pre-processing techniques resulted in an impressive accuracy precision and recall values. This analysis identified 20 important metabolites as essential components of the formula, which we predicted as natural antibiotics. In the final stage of our investigation, we verified our prediction by conducting a literature search for journal validation or by analyzing the structural similarity with known antibiotics using asymmetric similarity.

Robust optimization strategies for contour uncertainties in online adaptive radiation therapy

Objective. Online adaptive radiation therapy requires fast and automated contouring of daily scans for treatment plan re-optimization. However, automated contouring is imperfect and introduces contour uncertainties. This work aims at developing and comparing robust optimization strategies accounting for such uncertainties. Approach. A deep-learning method was used to predict the uncertainty of deformable image registration, and to generate a finite set of daily contour samples. Ten optimization strategies were compared: two baseline methods, five methods that convert contour samples into voxel-wise probabilities, and three methods accounting explicitly for contour samples as scenarios in robust optimization. Target coverage and organ-at-risk (OAR) sparing were evaluated robustly for simplified proton therapy plans for five head-and-neck cancer patients. Results. We found that explicitly including target contour uncertainty in robust optimization provides robust target coverage with better OAR sparing than the baseline methods, without increasing the optimization time. Although OAR doses first increased when increasing target robustness, this effect could be prevented by additionally including robustness to OAR contour uncertainty. Compared to the probability-based methods, the scenario-based methods spared the OARs more, but increased integral dose and required more computation time. Significance. This work proposed efficient and beneficial strategies to mitigate contour uncertainty in treatment plan optimization. This facilitates the adoption of automatic contouring in online adaptive radiation therapy and, more generally, enables mitigation also of other sources of contour uncertainty in treatment planning.

Flow unit classification and characterization with emphasis on the clustering methods: a case study in a highly heterogeneous carbonate reservoir, eastern margin of Dezful Embayment, SW Iran

Previous attempts to classify flow units in Iranian carbonate reservoirs, based on porosity and permeability, have faced challenges in correlating the rock's pore size distribution with the capillary pressure profile. The innovation of this study highlights the role of clustering techniques, such as Discrete Rock Type, Probability, Global Hydraulic Element, and Winland's Standard Chart in enhancing the reservoir's rock categorization. These techniques are integrated with established flow unit classification methods. They include Lucia, FZI, FZI*, Winland R35, and the improved stratigraphic modified Lorenz plot. The research accurately links diverse pore geometries to characteristic capillary pressure profiles, addressing heterogeneity in intricate reservoirs. The findings indicate that clustering methods can identify specific flow units, but do not significantly improve their classification. The effectiveness of these techniques varies depending on the flow unit classification method employed. For instance, probability-based methods yield surpassing results for low-porosity rocks when utilizing the FZI* approach. The discrete technique generates the highest number of flow unit classes but provides the worst result. Not all clustering techniques reveal discernible advantages when integrated with the FZI method. In the second part, the study creatively suggests that rock classification can be achieved by concurrently clustering irreducible water saturation (SWIR) and porosity in unsuccessful flow unit delineation cases. The SWIR log was estimated by establishing a smart correlation between porosity and SWIR in the pay zone, where water saturation and SWIR match. Then, the estimated saturation was dispersed throughout the reservoir. Subsequently, the neural network technique was employed to cluster and propagate the three finalized flow units. This methodology is an effective recommendation when conventional flow unit methods fail. The study also investigates influential factors causing the failure of flow unit classification methods, including pore geometry, oil wettability, and saturation in heterogeneous reservoirs.

A new cosine-based approach for modelling the time-to-event phenomena in sports and engineering sectors

Data analysis through probability-based methods plays an important role in decision-making in all spheres of life. Probability-based approaches lead to productive decision-making under uncertainty. To date, several probability distributions have been formulated that are characterized in different ways and successfully applied in various fields. Most of these probability distributions are studied by adding many additional parameters, typically from one to eight. There is the very least number of available distributional methodologies that are devised without inducing new parameters. In this integrated research piece of work, we provide a new cosine-adaptive method that is developed without considering any new parameters. We call the proposed method the weighted cosine-K family of distributions. Focusing on the proposed method, a new probability model is developed that is named the weighted cosine-Weibull (WC-Weibull) distribution. Some quartile-based characteristics along with estimators of the WC-Weibull distribution are derived. For the parameters of the WC-Weibull distribution, the point estimators are derived. The assessment of point estimators of the WC-Weibull distribution's parameters is provided through a simulation study. Finally, the utility of the WC-Weibull distribution is demonstrated by considering time-to-event phenomena in sports and the engineering sectors. Considering four statistical criteria, it has been shown that the WC-Weibull distribution is an efficient and optimal model for time-related facts (or events) in sports and engineering fields.

Class-relevant feature density estimator for open set fault classification of industrial equipment using vibration signals

Recently, the application of vibration signals to the fault diagnosis of industrial equipment has attracted increased attention. Open set recognition allows deep networks to detect unknown faults while maintaining a high-classification accuracy for known faults, which is necessary for practical industrial applications. Existing distance- and probability-based open set recognition methods show significant potential in handling this challenging task. In this study, we combine a deep backbone network and probabilistic model to construct a class-relevant feature density estimator (CRFDE) that integrates the advantages of distance- and probability-based methods. Specifically, a loss function is developed to enable the backbone network to extract the class-relevant features and probabilistic model to estimate them, where the employed probabilistic model is improved from variational auto-encoder (VAE) to class-conditional VAE (CCVAE) that can estimate the class-conditional data densities. The output class-relevant feature densities of CCVAE are treated as special distances to determine the class belongingness of the input samples. We validated the proposed CRFDE on two public motor bearing datasets and one laboratory gas-insulated switchgear vibration signal dataset. The results revealed that the proposed CRFDE achieved higher F1 scores than the existing methods.

DeepFLR facilitates false localization rate control in phosphoproteomics

Protein phosphorylation is a post-translational modification crucial for many cellular processes and protein functions. Accurate identification and quantification of protein phosphosites at the proteome-wide level are challenging, not least because efficient tools for protein phosphosite false localization rate (FLR) control are lacking. Here, we propose DeepFLR, a deep learning-based framework for controlling the FLR in phosphoproteomics. DeepFLR includes a phosphopeptide tandem mass spectrum (MS/MS) prediction module based on deep learning and an FLR assessment module based on a target-decoy approach. DeepFLR improves the accuracy of phosphopeptide MS/MS prediction compared to existing tools. Furthermore, DeepFLR estimates FLR accurately for both synthetic and biological datasets, and localizes more phosphosites than probability-based methods. DeepFLR is compatible with data from different organisms, instruments types, and both data-dependent and data-independent acquisition approaches, thus enabling FLR estimation for a broad range of phosphoproteomics experiments.

Optimization on building combined cooling, heating, and power system considering load uncertainty based on scenario generation method and two-stage stochastic programming

The energy demand uncertainty is a complex issue for design tasks of building combined cooling, heating, and power (BCHP) systems. To overcome this problem, the characteristics of energy demand uncertainty need to be precisely described by generating massive scenarios. Here, the aim of the research is to study the effects of scenario generation methods on BCHP optimization considering energy demand uncertainty. Four scenario generation methods were compared and studied—two conventional probability-based methods using Monte-Carlo simulation and the Latin hypercube method to sample the basic scenario, as well as simulation-based methods based on building performance simulation. Two optimization models, including an independent scenario optimization (ISO) model and a two-stage stochastic programming (TSSP) model, were used to minimize the total annual cost of the BCHP system. Dynamic time warping (DTW) was applied for investigating differences between scenario generation methods. The results show that Monte Carlo building performance simulation (MCBPS) is the most similar to the basic scenario. The system cost obtained by the uncertainty based optimization method is 1.5% - 3.6% higher than that of the deterministic optimization method. This is mainly due to the capacity increase of peak shaving equipment. From the perspective of anti risk, relatively higher internal combustion engine (ICE) capacity has stronger anti risk ability, which can effectively reduce the total cost of the system. This study can provide a theoretical reference for architectural design and operation research based on uncertainty.

Acute stress, worry, and impairment in health care and non-health care essential workers during the COVID-19 pandemic.

Health care and non-health care essential workers working in face-to-face interactions during the coronavirus disease 2019 (COVID-19) pandemic may be vulnerable to psychosocial distress. Limited empirical research on COVID-19-related psychosocial outcomes has utilized probability-based samples including both health care and non-health care essential workers. We surveyed a sample of 1,821 United States self-identified essential workers, collected using probability-based methods, working in face-to-face interactions during the early phase of the COVID-19 outbreak (March 18, 2020 through April 18, 2020), in three consecutive 10-day cohorts. We assessed acute stress, health-related worries, and functional impairment. Demographics, secondary stressors (lack of childcare or health care, lost wages), and pre-COVID-19 mental and physical health were examined as predictors of psychological outcomes. Acute stress (β = .08, p = .001), health-related worries (β = .09, p = .001), and functional impairment (β = .05, p = .034) increased over time in the early weeks of the outbreak. Health care essential workers reported lower functional impairment (β = -.06, p = .009) and acute stress (β = -.06, p = .015) compared with non-health care essential workers. Across the sample, prior mental and physical health ailments, inability to obtain health care, lost wages, younger age, female gender, and Hispanic ethnicity were associated with acute stress (βs = -.14 to .15, ps ≤ .001), health-related worries (βs = -.09 to .14, ps ≤ .001), and functional impairment (βs = -.08 to .16, ps ≤ .006). Lack of childcare (β = .09, p < .001) was positively associated with acute stress. Non-health care essential workers may be vulnerable to negative psychosocial outcomes. Targeted training and support may help facilitate coping with the effects of working in-person during the ongoing COVID-19 pandemic. Findings may help inform intervention efforts, critical as COVID-19 becomes endemic and society must learn to live with its evolving variants. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

Compton coincidence in silicon photon-counting CT detectors.

Purpose: Compton interactions amount to a significant fraction of the registered counts in a silicon detector. In a Compton interaction, only a part of the photon energy is deposited and a single incident photon can result in multiple counts unless tungsten shielding is used. Deep silicon has proved to be a competitive material for photon-counting CT detectors, but to improve the performance further, one possibility is to use coincidence techniques to identify Compton-scattered photons and reconstruct their incident energies. Approach: In a detector with no tungsten shielding, incident photons can interact through a series of interactions. Based on the position and energy of each interaction, probability-based methods can be used to estimate the incident photon energy. Here, we present a maximum likelihood estimation framework along with an alternative method to estimate the incident photon energy and position in a silicon detector. Results: Assuming one incident photon per time frame, we show that the incident photon energy can be estimated with a mean error of and an RMS error of for a realistic case in which we assume a detector with limited energy and spatial resolution. The interaction position was estimated with a mean error of in direction and in direction. Corresponding RMS errors of and were achieved in and , respectively. Conclusions: The presented results show the potential of using probability-based methods to improve the performance of silicon detectors for CT.

Efficient approaches for large-scale GWAS with genotype uncertainty.

Association studies using genetic data from SNP-chip-based imputation or low-depth sequencing data provide a cost-efficient design for large-scale association studies. We explore methods for performing association studies applicable to such genetic data and investigate how using different priors when estimating genotype probabilities affects the association results. Our proposed method, ANGSD-asso’s latent model, models the unobserved genotype as a latent variable in a generalized linear model framework. The software is implemented in C/C++ and can be run multi-threaded. ANGSD-asso is based on genotype probabilities, which can be estimated using either the sample allele frequency or the individual allele frequencies as a prior. We explore through simulations how genotype probability-based methods compare with using genetic dosages. Our simulations show that in a structured population using the individual allele frequency prior has better power than the sample allele frequency. In scenarios with sequencing depth and phenotype correlation ANGSD-asso’s latent model has higher statistical power and less bias than using dosages. Adding additional covariates to the linear model of ANGSD-asso’s latent model has higher statistical power and less bias than other methods that accommodate genotype uncertainty, while also being much faster. This is shown with imputed data from UK Biobank and simulations.

Seventh-Grade Students’ Levels of Associating Science Courses with Real Life

This study aims to reveal the extent to which secondary seventh-grade students can associate what they learn in science courses with real life. The study was conducted with a descriptive method after selecting participants through probability-based sampling methods, namely, stratified sampling and cluster sampling. The research was carried out with 274 seventh-graders from 12 secondary schools in the central districts of Ankara province during the 2017-2018 academic year. The students’ levels of associating science with real life were studied from two perspectives: proposing solutions to real life problems by drawing on scientific principles and identifying the scientific principles applicable in solving related life problems. To properly assess these measures, two tests were developed by the researcher and were used. These tests consist of open-ended problem scenarios and are called the “Test of Life in Science” and the “Test of Science in Life”. The data analysis results revealed that the students cannot sufficiently associate science courses with life, as they cannot propose solutions for life problems employing scientific principles and they also cannot elicit the principles applied in solving life problems related to science. Moreover, the students’ levels of associating science courses with life significantly vary depending on the students’ socioeconomic status, access to supplementary science courses, domestic Internet access, and out-of-school experience of science learning. The study results seem significant, as they imply the need for necessary arrangements to educate students as individuals who are able to do the following: solve real-life problems by using their scientific knowledge; make decisions and put forward useful outcomes for the good of society by using such knowledge from good science education; adapt easily to their community; and develop favorable consumption habits.

Reliability theory for microbial water quality and sustainability assessment

Microbial surface water contamination can disrupt critical ecosystem services such as recreation and drinking water supply. Prediction of water contamination and assessment of sustainability of water resources in the context of water quality are needed but are difficult to achieve – with challenges arising from the complexity of environmental systems, and stochastic variability of processes that drive contaminant fate and transport. In this paper we use reliability theory as a framework to address these issues. We define failure as exceedance of regulatory water contamination limits, and system components as reaches in the surface water network. We then methodically study the reliability of each component in the context of water quality, as well as the impact of individual components on overall water quality and sustainability. We obtain spatially distributed probability- and physics-based sustainability measures of reliability, vulnerability, resilience and the sustainability index. Finally, we use GIS as a platform to present these measures as geospatial products in an effort to foster public acceptance of probability-based methods in contaminant hydrology.

Abstract A30: Comparing mail-based and Internet population health assessment data: Survey implementation protocol

Abstract Purpose of the Study: Traditionally, population-based health assessments have been conducted through resource intensive approaches such as in-person interviews and mail-based surveys. Newer information technologies have been developed for collecting survey data electronically (i.e., e-cohort approaches), which can be leveraged to recruit participants and collect survey data more efficiently than traditional studies. Funded by a National Cancer Institute P30 Supplement, the purpose of this study overall is to assess health practices and improve cancer prevention and control outreach within our cancer center’s catchment area. The purpose of this report is to describe procedures of this ongoing project collecting survey data through traditional and electronic methods. Methods: This study will administer a survey using two approaches to population health assessment. The first survey cohort will be obtained via a mail-based survey using traditional probability-based sampling methods. A second cohort will be recruited and assessed through technology-enabled methods. This e-cohort will be contacted using an Internet and mobile phone application adapted from the platform developed by Vibrent Health for the National Institutes of Health All of Us study. In addition to the survey responses, data from wearable devices and health service records data will be collected from participants of the e-cohort. Results: This protocol report will describe implementation procedures of an epidemiologic research study employing both conventional and novel approaches to survey administration. Upon completion of this study, results will provide metrics for comparing the methodologies, such as response rates and respondent characteristics (e.g., socio-demographics, medical history, and lifestyle behaviors) within and between each approach. Conclusions: Ultimately, results of this study will aid the development of future epidemiologic research studies by providing practical guidance for public health and survey researchers regarding traditional and electronic survey applications, and perhaps, novel insights from the complementary approach of using both survey research methodologies in tandem. Citation Format: Bernard F. Fuemmler, Carrie A. Miller, David Wheeler, Bassam Dahman, Tamas Gal, Sunny Kim, Bonnie Morris, Yaou Sheng, Gordon Ginder. Comparing mail-based and Internet population health assessment data: Survey implementation protocol [abstract]. In: Proceedings of the AACR Special Conference on Modernizing Population Sciences in the Digital Age; 2019 Feb 19-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2020;29(9 Suppl):Abstract nr A30.

Challenges and Possibilities of Overtaking Strategies for Autonomous Vehicles

This paper present three distinct probability-based methods for decision making and trajectory planning layers of overtaking maneuvering functionality for autonomous vehicles. The computation time of the proposed decision-making algorithms may be high, because the number of describing parameters of the traffic situations may vary in a high range. The presented clustering-based, graph-based and dynamic-based methods differ in the complexity of their computation algorithms. Since the decision-making process may require considerable online computation effort, a neural-network-based approach is presented for implementation purposes.

ASSESSMENT OF THE RELIABILITY OF THE REINFORCED CONCRETE LAYER IN STRUCTURES

An urgency to use probability-based methods and probabilistic reliability indices in long-term quality assessments and predictions of reinforced concrete products and structures is under consideration. Carbonation nature and depth of protective concrete covers of reinforcement bars and stirrups are analysed. Effect of water-cement ratio on the carbonation rate of protective concrete covers is discussed. Analyzes of reliability of structures reinforcement with composite as spiral (transverse) usage and influence. Presented applied equations and the calculation example demonstrate a simplicity and necessities of probability-based methods.

A Confidence-Guided Evaluation for Log Parsers Inner Quality

Logs faithfully record application behaviors and system states. Log parsing converts unstructured log messages into structured event templates by extracting the constant portion of raw logs. Log parsing is a prerequisite for further log analysis such as usage analysis, anomaly detection, performance modeling, and failure diagnosis. When processing logs with varied length, log parsing suffers from accuracy decreasing or the over-fitting problem. In addition, traditional probability-based accuracy assessment methods are ineffective in assessing log parsing inner quality, especially in understanding the reason about accuracy declining caused by varied length logs. In this paper we present a p_value-guided inner quality assessment on multiple log parsing algorithms. This method uses conformal evaluation to gain a deep insight of log parser quality. In this method, we choose the string edit distance algorithm as underlying non-conformity measure for conformal evaluation. We introduce two quality indicators to evaluate log parsers: credibility and confidence. The credibility reflects how conformal a log message to a event template generated by a log parser whereas the confidence reflects how non-conformal this log message to all other event templates. In order to demonstrate the inherent difference among different log parsers, we display the distribution of credibility and confidence of each prediction on tSNE 2D space. In the experiment, we evaluate 13 log parsers on different datasets. The results show that our approach could effectively demonstrate the inherent quality of log parsers and recognize variable-length problem compared to traditional confusion matrix based metrics.

DMGAN: Discriminative Metric-based Generative Adversarial Networks

With the proposed of Generative Adversarial Networks (GANs), the generative adversarial models have been extensively studied in recent years. Although probability-based methods have achieved remarkable results in image synthesis tasks, there are still some unsolved challenges that are difficult to overcome. In this paper, we propose a novel model, called Discriminative Metric-based Generative Adversarial Networks (DMGANs), for generating real-like samples from the perspective of deep metric learning. To be specific, the generator is trained to generate realistic samples by reducing the distance between real and generated samples. Instead of outputting probability, the discriminator in our model is conducted as a feature extractor, which is well constrained by introducing a combination of identity preserving loss and discriminative loss. Meanwhile, to reduce the identity preserving loss, we calculate the distance between samples and their corresponding center and update these centers during training to improve the stability of our model. In addition, a data-dependent strategy of weight adaption is proposed to further improve the quality of generated samples. Experiments on several datasets illustrate the potential of our model.

Transcranial magnetic stimulation induced early silent period and rebound activity re-examined.

Despite being widely studied, the underlying mechanisms of transcranial magnetic brain stimulation (TMS) induced motor evoked potential (MEP), early cortical silent period (CSP) and rebound activity are not fully understood. Our aim is to better characterize these phenomena by combining various analysis tools on firing motor units. Responses of 29 tibialis anterior (TA) and 8 abductor pollicis brevis (APB) motor units to TMS pulses were studied using discharge rate and probability-based tools to illustrate the profile of the synaptic potentials as they develop on motoneurons in 24 healthy volunteers. According to probability-based methods, TMS pulse produces a short-latency MEP which is immediately followed by CSP that terminates at rebound activity. Discharge rate analysis, however, revealed not three, but just two events with distinct time courses; a long-lasting excitatory period (71.2 ± 9.0 ms for TA and 42.1 ± 11.2 ms for APB) and a long-latency inhibitory period with duration of 57.9 ± 9.5 ms for TA and 67.3 ± 13.8 ms for APB. We propose that part of the CSP may relate to the falling phase of net excitatory postsynaptic potential induced by TMS. Rebound activity, on the other hand, may represent tendon organ inhibition induced by MEP activated soleus contraction and/or long-latency intracortical inhibition. Due to generation of field potentials when high intensity TMS is used, this study is limited to investigate the events evoked by low intensity TMS only and does not provide information about later parts of much longer CSPs induced by high intensity TMS. Adding discharge rate analysis contributes to obtain a more accurate picture about the characteristics of TMS-induced events. These results have implications for interpreting motor responses following TMS for diagnosis and overseeing recovery from various neurological conditions.

Workshop: Reaching out to engaging the risk groups: online respondent-driven methods for public health

Abstract Hard-to-reach populations (i.e. those stigmatised, marginalised, underrepresented, or otherwise disadvantaged) such as men who have sex with men and immigrants are at higher risk for infectious diseases. Reaching these populations, studying their behaviour and/or testing individuals for infectious diseases is essential for developing effective prevention programmes and disease surveillance. These populations, however, lack sampling frames making it difficult to collect representative quantitative data using common probability-based sampling methods. Respondent-driven sampling (RDS), a variant of snowball sampling, is an effective method to recruit these populations and to make unbiased population estimates using a statistical model. RDS starts with recruiting a convenience sample of the target population (so-called “seeds”). These seeds are then asked to recruit a number of other eligible individuals of their social network. This process continues which leads to chains of recruitment, with several waves of recruits. The process of respondent-driven recruitment is very similar to the way infectious diseases such as influenza and mumps transmit through populations. This provides opportunities to use the method with a different aim: the detection of cases within networks. Finding infectious cases is an essential element for prevention of further spread in the population and individual health consequences. Essential as it is to public health, conventional contact tracing is a rather timely, costly and, up to a certain degree, really frustrating activity. Studying and making use of social networks may help to understand and control the spread of infectious diseases transmitted via direct contact. These diseases do not spread at random through a population, but follow the underlying patterns of contact networks. This entails that cases tend to cluster by time and space, and their contact persons are at a higher risk for infection. Same as with RDS, respondent-driven detection (RDD) starts with individuals being asked to recruit relevant contact persons from their network. These contact persons are then asked to do the same, resulting in successive waves of contact persons. A case is reached through contact with a known case, similar to pathogens spreading through these contact relationships. RDD may therefore enhance conventional contact tracing, providing further insight in the extent of outbreaks, in a quick and less laborious manner for public health professionals. Using three examples from public health practice, this workshop provides participants insights in the methodology of online respondent-driven methods (RDS and RDD), how these provide behavioural and epidemiological knowledge on networks and the spread of infectious diseases, and highlights pre-requisites for successful implementation in practice. Lastly, an interactive discussion will be held with the audience on how attendees can apply these methods for their own public health challenges. Key messages RDS is used to sample hard-to-reach populations to collect their social, sexual and behavioural information, and to make unbiased population estimates. RDD is used to detect infectious cases or clusters of disease.

Interpretation of intelligence in CNN-pooling processes: a methodological survey

The convolutional neural network architecture has different components like convolution and pooling. The pooling is crucial component placed after the convolution layer. It plays a vital role in visual recognition, detection and segmentation course to overcome the concerns like overfitting, computation time and recognition accuracy. The elementary pooling process involves down sampling of feature map by piercing into subregions. This piercing and down sampling is defined by the pooling hyperparameters, viz. stride and filter size. This down sampling process discards the irrelevant information and picks the defined global feature. The generally used global feature selection methods are average and max pooling. These methods decline, when the main element has higher or lesser intensity than the nonsignificant element. It also suffers with locus and order of nominated global feature, hence not suitable for every situation. The pooling variants are proposed by numerous researchers to overcome concern. This article presents the state of the art on selection of global feature for pooling process mainly based on four categories such as value, probability, rank and transformed domain. The value and probability-based methods use the criteria such as the way of down sampling, size of kernel, input output feature map, location of pooling, number stages and random selection based on probability value. The rank-based methods assign the rank and weight to activation; the feature is selected based on the defined criteria. The transformed domain pooling methods transform the image to other domains such as wavelet, frequency for pooling the feature.