International Journal of Computing Science and Mathematics (7 papers in press)

Regular Issues

• Stability and bifurcation study of interaction in the vermifiltration phase between predators and prey  
  by Madhan Kumar, Mullai Murugappan 
  Abstract: The issue of waste water disposal that poses a major challenge especially in the industrial sector is discussed in this article. Vermifiltration is used to transform toxic waste to non-hazardous waste. Our focus is on the survival of living organisms which is involved in the process of vermifiltration. We formulate and build a prey predator model with stage structure for the predator population. Model equilibria are observed and studied. The proposed model is expanded by incorporating time delays in the model. The global stability (with and without delay) of the model is discussed in detail. Our findings show that the increase in the density mortality rate of the predator maintains the equilibrium to a certain degree and Hopf bifurcation occurs in the model beyond this. The system exhibits oscillatory behavior when the gestation time delay reaches the threshold level. Further, computational simulations are demonstrated and biological explanations are provided.
  Keywords: vermifiltration; prey-predator; time delay; equilibria; stability; Hopf bifurcation.
  
  
• Incorporation of Question Segregation Procedure in Visual Question Answering Models  
  by Souvik Chowdhury, Badal Soni, Doli Phukan 
  Abstract: There are various open issues in visual question answering (VQA). One of them is sometimes a model can predict Yes or No as an answer which is not relatable to the question and requires a descriptive answer and vice versa. To solve this issue in the VQA domain, in this paper, a question segregation (QS) technique is incorporated to classify the questions into three types (Yes/No, Other and Number). Then we successfully incorporated this technique with wo of the VQA models, stacked attention networks (SAN) and modular co-attention network (MCAN). We evaluate the performance of the QS and SAN models on two datasets VQA v.2 and CLEVR. We also studied and analysed the impact of Question Segregation on the performance of these two models on different datasets.
  Keywords: Visual Question Answering; Machine Learning; Deep Learning; CNN;LSTM.
  DOI: 10.1504/IJCSM.2023.10058564
   
  
• A Spark-based Parallel Genetic Algorithm for Bayesian Network Structure Learning  
  by Naixin Wu 
  Abstract: The Bayesian network structure learning (BNSL) algorithm based on genetic algorithm (GA) has the problem of long search time and being prone to falling into local optima. When the sampling data is large, the single machine BNSL algorithm cannot obtain the BN structure within a limited time. To address this issue, this paper proposes a parallel BNSL algorithm based on the Spark framework with GA (PGA-BN). The three main stages of the proposed PGA-BN are population initialization, BIC score calculation, and evolution operators, which are all designed in parallel on each partition to accelerate based on Spark. The experiments are studied on two typical BN datasets with different sample sizes to evaluate the parallel performance of the PGA-BN algorithm. Experimental results showed that the PGA-BN is significantly faster than its single-machine version with the satisfied accuracy.
  Keywords: Bayesian networks; structure learning; genetic algorithm; parallel.
  DOI: 10.1504/IJCSM.2023.10061827
   
  
• Zipper Quintic Fractal Interpolation Function for Curve Fitting  
  by Sneha ., Kuldip Katiyar 
  Abstract: In this paper, we introduce a class of novel C^2-zipper rational quintic fractal interpolation functions (Zipper-RQFIF) with variable scalings, in the form of rational type which has a quintic polynomial in the numerator and a quadratic polynomial in the denominator with three shape control parameters. We restrict the scaling functions and shape control parameters so that the proposed Zipper-RQFIF is positive, when the given data set is positive. Using this sufficient condition, some numerical examples of positive Zipper-RQFIF are presented to support our theory. This paper approaches the zipper rational quintic fractal interpolation problem as a generalisation of both quintic fractal and affine zipper fractal interpolants which shows more versatility and flexibility than classical and fractal interpolation functions (FIFs).
  Keywords: Zipper; Zipper Fractal Interpolation Function (ZFIF); Positivity; Rational Quintic Fractal Interpolation Function (RQFIF); Iterated function system (IFS).
  DOI: 10.1504/IJCSM.2024.10062995
   
  
• A Separable Convolutional Neural Network for Vehicle Type Recognition  
  by Baili Zhang, Yansu Wang 
  Abstract: The traditional vehicle type recognition algorithm has a low image recognition rate for various vehicle types on diverse road conditions and is prone to being affected by shooting distance, light intensity, and weather. To address these problems, a new separate convolutional neural network structure was proposed to automatically classify the images of different vehicle types based on the deep learning TensorFlow framework and the classical GoogLeNet-based network model. Experimental results on the data sets of BIT-Vehicle and Cars-196 show that compared with the traditional HOG_BP algorithm and convolutional neural network model, the decomposed convolutional neural network has a higher recognition rate for the same difficult vehicle images, and its average accuracy rate reaches 96.30%. In addition, the adjustment of hyperparameters in the network ensures that the parameters such as weight and bias amount are more efficient and reasonable when constantly updated.
  Keywords: Vehicle type recognition; GoogLeNet network; TensorFlow framework; Separate convolutional neural networks.
  DOI: 10.1504/IJCSM.2024.10064502
   
  
• Study on temperature distribution characteristics of fuel pools' fire in core engine cavity based on FDS  
  by Guanbing Cheng, Xiwei LIU, Zhangyuan CHEN 
  Abstract: Temperature distribution behaviours in aviation engine is important during design of engine's fire protection system. The present paper examined diffusion characteristics of pools' fuel fire in core engine cavity. Physical and numerical models of three pools' fire were established. After the grid size was validated, variations of flame temperature and its shapes were examined. The results show that the temperature variation of pools' fire undergoes increasing and steady stages. In cavity left area, the left single pool fire temperature is higher than that in double pools. In cavity middle area, the double pools fire temperature is higher than those in both single pools as the smoke gathers at the casings. In cavity right area, the right single pool temperature is higher than that in double pools. The flame shapes of the double pools undergo floating, inclination, partially and completely merging stages. The single pool flame originally floats, then slightly inclines.
  Keywords: Turbofan; core engine cavity; double fuel pools; FDS; temperature distribution; flame shape.
  DOI: 10.1504/IJCSM.2024.10064781
   
  
• Architecture Generation for Multi-objective Neural Architecture Search  
  by Songyi Xiao, Wenjun Wang 
  Abstract: Architecture generation in neural architecture search (NAS) has garnered significant attention due to its efficient architecture generation. Learning the architecture representation through unsupervised learning and creating a latent space simplifies the prediction process of predictors, enhancing efficiency in architecture search. Nevertheless, many NAS approaches prioritize identifying architectures based solely on accuracy, overlooking architectural complexity. This paper introduces a multi-objective neural architecture approach that combines a multi-objective evolutionary algorithm and a generative model. The proposed approach tackles the challenge by regularizing the latent space and achieving a balance between architecture accuracy and complexity. Moreover, the introduction of ranking error aids in gradually regulating the generative model, simplifying the identification of architectural representations. In addition, a multi-objective evolutionary algorithm, constructed based on a reference point, is utilized to uphold the quality of architectures. The experiment results demonstrate the effectiveness of AG-MONAS in selecting architectures that strike a trade-off between accuracy and complexity.
  Keywords: Neural architecture search; multi-objective problem; ranker; generative model.
  DOI: 10.1504/IJCSM.2024.10064958