International Journal of Bio-Inspired Computation (8 papers in press)

Regular Issues

• Swap Operator based Grey Wolf Optimizer for Materialized View Selection Problem  
  by Amit Kumar 
  Abstract: Grey wolf optimiser (GWO) is a novel biological intelligence algorithm that primarily simulates the natural leadership structure and hunting behaviour of grey wolves. It has the ability to decrease the operating time for higher dimensional data by partitioning the large, complicated problems into smaller sub problems and distributing the subsets of operations to each agent. Therefore, it is widely employed in a variety of time-consuming tasks, including NP-hard problems. Materialised view selection (MVS) is an NP-hard discrete optimisation problem in the design process of a data warehouse that significantly speeds up query processing. Therefore, in this paper, GWO has been discretised using swap operators and swap sequence operators to address the MVS problem. Experimentally, it is observed that the proposed swap operator-based grey wolf optimiser (SOGWO) algorithm selects better quality views for materialisation than those selected using well-known metaheuristic algorithms over a number of view selection problem instances.
  Keywords: grey wolf optimiser; GWO; data warehouse; materialised view; view selection; swap operator.
  DOI: 10.1504/IJBIC.2023.10062102
   
  
• Integrating Convolution and Transformer for Enhanced Diabetic Retinopathy Detection  
  by Xinrong Cao, Jie Lin, Xiao-Zhi Gao, Zuoyong Li 
  Abstract: Diabetic retinopathy (DR) is a common diabetes complication that can cause irreversible blindness. Deep learning models have been developed to automatically classify the severity of retinopathy. However, these methods face challenges like a lack of long-range connections, weak interactions between images, and mismatches between lesion details and receptive fields, leading to accuracy issues. In our research, we propose a deep learning model with three main aspects. Firstly, a transformer structure is incorporated into a convolutional neural network to effectively utilise both local and long-range information. Secondly, the disease details are aggregated from multiple images before applying self-attention to improve inter-image interactions and reduce overfitting. Lastly, an attention-based approach is proposed to filter information from different stages of feature maps and adaptively capture lesion-related details. Our experiments achieved a 5-class accuracy of 85.96% on the APTOS dataset and a 2-class accuracy of 95.33% on the Messidor dataset, surpassing recent methods.
  Keywords: diabetic retinopathy; DR; convolutional neural network; transformer; cross attention; deep feature aggregation.
  DOI: 10.1504/IJBIC.2023.10062454
   
  
• Intelligent MPPT for Photovoltaic Panels on Grid-connected Inverter System using Hybrid Meta-Heuristic Algorithm  
  by Sebi N.P.  
  Abstract: This research work presents a novel hybrid meta-heuristic algorithm-based MPPT controller for efficiently tracking the maximum power under all weather conditions. The experimentation was conducted with the connection of direct current (DC) to alternating current (AC) single-phase full bridge inverter and optimised fractional order proportional integral derivative (FoPID) controller. The integrated algorithm named crow electric fish search optimisation (CEFSO) is utilised for optimising the global maximum power point (GMPP) of the MPPT. The major scope of optimising this constraint is to attain higher energy from the PV system. The proposed CEFSO-MPPT shows superior efficiency regarding faster convergence at the GMPP and MPPT. The suggested MPPT technique has shown its performance concerning higher efficiency, lower computational burden, and faster MPP tracking while analysing the existing models.
  Keywords: photovoltaic panels on grid-connected inverter system; DC-AC single-phase full-bridge inverter; maximum power point tracking; MPPT; optimised fractional proportional integral derivative controller.
  DOI: 10.1504/IJBIC.2023.10062802
   
  
• Determinate Node Selection for Semi-supervised Classification Oriented Graph Convolutional Networks  
  by Yao Xiao, Ji Xu, Yang Jing, Li Shaobo, Guoying Wang 
  Abstract: Graph convolutional networks (GCNs) have been proved successful in the field of semi-supervised node classification by extracting structural information from graph data. However, the random selection of labelled nodes used in GCNs may lead to unstable generalisation performance of GCNs. In this paper, we propose an efficient method for the deterministic selection of labelled nodes: the determinate node selection (DNS) algorithm. The DNS algorithm identifies two categories of representative nodes in the graph through structural analysis of the leading tree information granules: typical nodes and divergent nodes. These labelled nodes are selected by exploring the structure of the graph and determining the ability of the nodes to represent the distribution of data within the graph. The DNS algorithm can be applied quite simply on GCNs, and a wide range of semi-supervised graph neural network models for node classification tasks. Through extensive experimentation, we have demonstrated that the incorporation of the DNS algorithm leads to a remarkable improvement in the average accuracy of the model and a significant decrease in the standard deviation simultaneously, as compared to the vanilla method without a DNS module.
  Keywords: graph convolutional networks; granular computing; semi-supervised learning; node classification.
  DOI: 10.1504/IJBIC.2024.10062817
   
  
• A Holistic Approach to Early Congenital Heart Disease Detection in Rural Neonates: Bridging the Gap in Postnatal Care  
  by Rishika Anand, S.R.N. Reddy, Dinesh Kumar Yadav 
  Abstract: This research seeks to combat the alarming neonatal mortality rates attributed to undiagnosed congenital heart disease (CHD) in underserved rural regions, which often lack the necessary medical equipment and treatments. To address this pressing issue, we propose the development of an early detection system capable of identifying CHD in infants, thus facilitating timely intervention and enhancing postnatal care. Our study advocates an integrated approach that comprehensively analyses various vital parameters, including blood pressure, ECG, SpO2, body temperature, and heart rate in newborns, ensuring the accurate detection of CHD. The primary objective is to enhance the support and guidance provided to families in rural areas concerning postnatal care, prognosis, and treatment strategies. To gauge the effectiveness of our approach, we will compare it with existing techniques and evaluate its precision. The outcomes of this research hold the potential to significantly diminish neonatal mortality rates in these underserved regions.
  Keywords: machine learning; deep learning; congenital heart disease; CHD; early detection.
  DOI: 10.1504/IJBIC.2024.10062935
   
  
• Fitness Function with Two Rankings Embedded in a Push and Pull Search Framework for Constrained Multi-objective Optimisation Problems  
  by Kangshun Li, Jiaxin Xu, Shumin Xie, Hui Wang 
  Abstract: Optimizing objectives and satisfying constraints present significant challenges in solving constrained multi-objective optimisation problems. In this paper, we propose an algorithm that incorporates the push-and-pull search framework and a two-ranking fitness function named ToR-PPS. The algorithm is divided into three stages: the push stage, transitional stage, and pull stage. In the push stage, the population is directed toward the unconstrained Pareto front, without consideration of constraints. In the transitional stage, a diversity expansion strategy is proposed to optimise the diversity of the population. In the pull stage, the fitness function with two rankings is utilised to pull the population toward the constrained Pareto front. Experiments are conducted to compare the algorithm with five state-of-the-art constrained multi-objective optimisation evolutionary algorithms on three benchmark suites. The results clearly illustrate the superiority and efficiency of the algorithm.
  Keywords: constrained multi-objective optimization problems; evolutionary algorithms; push and pull search framework; transitional stage; diversity.
  DOI: 10.1504/IJBIC.2024.10062987
   
  
• Research on micro defect recognition based on deep learning  
  by Donghong Yang, Yude He 
  Abstract: To improve the overall recognition accuracy for micro defects, magnetic tile defect images are taken as the research object, and a defect image recognition method based on deep learning is proposed. Using MobileNetV3 network as the basic model, the number of deep convolution and the number of channels are trimmed. Then, mish function is used to replace h-swish function as the activation function, and the training speed and recognition accuracy of the model are improved, thus realising efficient and accurate recognition of micro defect images. The simulation results show that compared with the recognition methods standard MobileNetV3 network and other classification models faster R-CNN and EfficientNet, the proposed method performs better in terms of accuracy and F1 value, reaching 99.59% and 98.75%, respectively. The proposed method can recognise the defect image more accurately, and has the advantages of high inference speed, low parameter quantity and low computational cost.
  Keywords: mage recognition; image segmentation; deep learning; MobileNetV3 network.
  DOI: 10.1504/IJBIC.2024.10063158
   
  
• Improved Harmonic Spectral Envelop Extraction for Singer Classification with Hybridised Model  
  by Balachandra Kumaraswamy  
  Abstract: The singing voice has an effect on humans with the addition of expressions, lyrics, and instruments. It is easier for human beings to distinguish the singing tone of voice from a specified auditory clip owing to an individual's perceptual tools and audible physiology. On the other, without human intervention, it is not simpler to identify non-vocal portions, vocal portions, feelings, and singers from the related signal owing to intrinsic complications. This proposed a new singer classification mechanism with four stages: pre-processing, vocal segmentation, feature extraction, and classification. Initially, first stage, an improved convolutional neural network (CNN) is deployed for the segmentation of the vocal part. Further, features like zero crossing rate (ZCR), Mel-frequency cepstral coefficients (MFCCs), vibration estimation and improved harmonic spectral envelope are derived to bidirectional gated recurrent unit (BI-GRU) and long short-term memory (LSTM). The results from LSTM and BI-GRU are median and the final result is attained.
  Keywords: singer classification; zero crossing rate; ZCR; convolutional neural network; improved CNN; bidirectional gated recurrent unit; BI-GRU; long short-term memory; LSTM.
  DOI: 10.1504/IJBIC.2023.10063534