arXiv subjects

Ashis Kumar Chakraborty

Publications and source records attributed to Ashis Kumar Chakraborty.

6 recordsLinked to original sources

arXiv2022

Estimating Software Reliability Using Size-biased Modelling

Software reliability estimation is one of the most active areas of research in software testing. Since time between failures (TBF) has often been challenging to record, software testing data are commonly recorded as test-case-wise in a discrete set up. We have developed a Bayesian generalised linear mixed model (GLMM) based on software testing detection data and a size-biased strategy which not only estimates the software reliability, but also estimates the total number of bugs present in the software. Our approach provides a flexible, unified modelling framework and can be adopted to various real-life situations. We have assessed the performance of our model via simulation study and found that each of the key parameters could be estimated with a satisfactory level of accuracy. We have also applied our model to two empirical software testing data sets. While there can be other fields of study for application of our model (e.g., hydrocarbon exploration), we anticipate that our novel modelling approach to estimate software reliability could be very useful for the users and can potentially be a key tool in the field of software reliability estimation.

stat.AP↗

arXiv2020

Bayesian Neural Tree Models for Nonparametric Regression

Frequentist and Bayesian methods differ in many aspects, but share some basic optimal properties. In real-life classification and regression problems, situations exist in which a model based on one of the methods is preferable based on some subjective criterion. Nonparametric classification and regression techniques, such as decision trees and neural networks, have frequentist (classification and regression trees (CART) and artificial neural networks) as well as Bayesian (Bayesian CART and Bayesian neural networks) approaches to learning from data. In this work, we present two hybrid models combining the Bayesian and frequentist versions of CART and neural networks, which we call the Bayesian neural tree (BNT) models. Both models exploit the architecture of decision trees and have lesser number of parameters to tune than advanced neural networks. Such models can simultaneously perform feature selection and prediction, are highly flexible, and generalize well in settings with a limited number of training observations. We study the consistency of the proposed models, and derive the optimal value of an important model parameter. We also provide illustrative examples using a wide variety of real-life regression data sets.

stat.ME↗

arXiv2019

Optimum Testing Time of Software using Size-Biased Concepts

Optimum software release time problem has been an interesting area of research for several decades now. We introduce here a new concept of size-biased modelling to solve for the optimum software release time. Bayesian approach is used to solve the problem. We also discuss about the applicability of the model for a specific data set, though we believe that the model is applicable to all kind of software reliability data collected in a discrete framework. It has applications in other fields like oil exploration also. Finally, we compare favourably our model with another similar model published recently. We also provide in this article some future possibilities of research work.

stat.ME↗

arXiv2018

A novel distribution-free hybrid regression model for manufacturing process efficiency improvement

This work is motivated by a particular problem of a modern paper manufacturing industry, in which maximum efficiency of the fiber-filler recovery process is desired. A lot of unwanted materials along with valuable fibers and fillers come out as a by-product of the paper manufacturing process and mostly goes as waste. The job of an efficient Krofta supracell is to separate the unwanted materials from the valuable ones so that fibers and fillers can be collected from the waste materials and reused in the manufacturing process. The efficiency of Krofta depends on several crucial process parameters and monitoring them is a difficult proposition. To solve this problem, we propose a novel hybridization of regression trees (RT) and artificial neural networks (ANN), hybrid RT-ANN model, to solve the problem of low recovery percentage of the supracell. This model is used to achieve the goal of improving supracell efficiency, viz., gain in percentage recovery. In addition, theoretical results for the universal consistency of the proposed model are given with the optimal value of a vital model parameter. Experimental findings show that the proposed hybrid RT-ANN model achieves higher accuracy in predicting Krofta recovery percentage than other conventional regression models for solving the Krofta efficiency problem. This work will help the paper manufacturing company to become environmentally friendly with minimal ecological damage and improved waste recovery.

stat.AP↗

arXiv2018

Superensemble Classifier for Improving Predictions in Imbalanced Datasets

Learning from an imbalanced dataset is a tricky proposition. Because these datasets are biased towards one class, most existing classifiers tend not to perform well on minority class examples. Conventional classifiers usually aim to optimize the overall accuracy without considering the relative distribution of each class. This article presents a superensemble classifier, to tackle and improve predictions in imbalanced classification problems, that maps Hellinger distance decision trees (HDDT) into radial basis function network (RBFN) framework. Regularity conditions for universal consistency and the idea of parameter optimization of the proposed model are provided. The proposed distribution-free model can be applied for feature selection cum imbalanced classification problems. We have also provided enough numerical evidence using various real-life data sets to assess the performance of the proposed model. Its effectiveness and competitiveness with respect to different state-of-the-art models are shown.

cs.LG↗

arXiv2018

A Nonparametric Ensemble Binary Classifier and its Statistical Properties

In this work, we propose an ensemble of classification trees (CT) and artificial neural networks (ANN). Several statistical properties including universal consistency and upper bound of an important parameter of the proposed classifier are shown. Numerical evidence is also provided using various real life data sets to assess the performance of the model. Our proposed nonparametric ensemble classifier doesn't suffer from the `curse of dimensionality' and can be used in a wide variety of feature selection cum classification problems. Performance of the proposed model is quite better when compared to many other state-of-the-art models used for similar situations.

math.ST↗