International Journal of Bio-Inspired Computation (18 papers in press)

Regular Issues

• 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
   
  
• Research on estimation of permeability coefficients in microbial geotechnical soils based on data-driven models  
  by Mayao Cheng, Linsheng Chen 
  Abstract: Microbial geotechnical soil permeability coefficient estimation prediction is extremely valuable for the development of soil engineering. The study proposes an integrated data-driven model combining three base learners, RVM, ANFIS and iTLBO-ELM, assigning corresponding weights to each base learner through the PLS integrated model combination model, and applying the model to the prediction of microbial geotechnical soil permeability coefficient estimation. The MAE of the proposed integrated model has lower values compared to the single model, but the two metrics MAPE and RMSE are not the lowest; however, the integrated model outperforms both iTLBO-ELM and RVM in terms of MAPE, and the estimated predictions of permeability coefficients for November data are better than those for May. For iTLBO-ELM and RVM, the MAPE of PLS decreased by 5.51% and 1.56% respectively in May and 3.46% and 1.24% respectively in October. The integrated data-driven model proposed in the study can effectively achieve the estimated prediction of microbial geotechnical soil permeability coefficients and facilitate the intelligent acquisition of engineering permeability coefficients.
  Keywords: data-driven models; microorganisms; geotechnical land; permeability coefficients.
  DOI: 10.1504/IJBIC.2024.10064267
   
  
• Improved Dirichlet Mixture Model Clustering Algorithm for Medical Data Anomaly detection  
  by Lili Wu, Majid Khan Majahar Ali, F.A.M. Peishan, Tian Ying, Tao Li 
  Abstract: In order to address the issue of identifying over-diagnosis and anomaly expenses in the healthcare service process, a local outlier mining clustering algorithm (ILOF-DPMM) is proposed by combining the clustering-based local outlier factor (CBLOF) algorithm with Dirichlet mixture model (DPMM). By extracting the patient's hospitalisation records from the medical record homepage, the influencing factors of hospitalisation costs for different disease types are classified, and the random forest method is used to reduce the feature dimension by disease type. The feature extraction and dimensionality reduction methods adopted by this algorithm effectively cluster medical insurance expense data. When calculating the LOF value of data, using a weighted calculation method based on the similarity of discrete and continuous features can more accurately detect abnormal data points in the data set, and has the ability to detect new data in real time, thus improving detection accuracy and efficiency.
  Keywords: over-diagnosis; anomaly expenses; anomaly detection; DPMM; CBLOF.
  DOI: 10.1504/IJBIC.2024.10064803
   
  
• Pheromone-Inspired Multiple Moving Targets Search Method for Swarm Unmanned Aerial Vehicles in Environments with Unknown Obstacles  
  by Mao Wang, Shaowu Zhou, Hongqiang Zhang, Lianghong Wu 
  Abstract: A multiple moving targets search problem for swarm UAVs in environments with unknown obstacles is studied. The search is divided into roaming search and collaborative search; the multitarget search algorithm consists of task allocation, roaming search, collaborative search and obstacle avoidance. To convert between collaborative search and roaming search, a distance-based dynamic task allocation strategy is proposed. A confidence area pheromone for roaming search is proposed to reduce the repeated search times conducted in the same areas. Probabilistic finite PSO is proposed to adapt to search for moving targets in collaborative search. Furthermore, a boundary scanning-based obstacle avoidance strategy is improved to achieve efficient obstacle avoidance for UAVs in a grid environment. Based on the above, a multiple moving-target search algorithm mode is constructed. This mode shows better performance than existing methods as verified through simulation experiments, and provides a helpful alternative in postdisaster search, and other search fields.
  Keywords: swarm unmanned aerial vehicles; multiple moving targets search; confidence area pheromone; probabilistic finite particle swarm optimisation; PFPSO.
  DOI: 10.1504/IJBIC.2024.10065092
   
  
• Methanol Price Prediction Method Based on Multimodal Fusion by Using CNN-GRU and Attention Mechanism  
  by Shuang Luo, Xuhui Zhu, Zhiwei Ni, Pingfan Xia, Liping Ni 
  Abstract: Considering that text is important to methanol price prediction, the text and the quantity information need to be fused for achieving high precision prediction. Hence, we put forward a novel methanol price forecasting approach ground on text-quantity multimodal fusion using CNN-GRU-attention mechanism network, named MFCGAM, which merges quantity information and text information obtained from research report, information and investor comments. Firstly, Word2Vec model is applied to process text, and the text-quantity dual channel based on CNN and GRU is established to extract text and quantity features respectively. Secondly, attention mechanism is employed to get text-quantity fused characteristics, which are used to predict methanol price. The experimental outcomes of three real datasets show that MFCGAM model obtains superior performance than other traditional models. Additionally, predictive ability of models can be improved by adding texts, and it is found that the results of short-term prediction are better than that those of long-term forecasting when using texts. It provides a very useful predictive tool for smart scheduling of coking intelligent plants.
  Keywords: methanal price prediction; multimodal fusion; attention mechanism; gated recurrent unit; GRU; convolutional neural network; CNN.
  DOI: 10.1504/IJBIC.2024.10065641
   
  
• Multi-Agent Reinforcement Learning based on Self-Satisfaction in Sparse Reward scenarios  
  by Baofu Fang, Dandan Tang, Zaijun Wang, Hao Wang 
  Abstract: To solve the problem of sparse reward in reinforcement learning in multi-agent environments, this paper proposes an emotional model of self-satisfaction based on the role of human emotions in decision-making. The self-satisfaction emotional model composed of thirst for knowledge, recognition, and psychological gap is used as the internal motivation, and an internal emotional reward is generated as an effective supplement to the external reward, to alleviate the problem of the sparse reward. Based on this model, a self-satisfaction-based multi-agent reinforcement learning algorithm is proposed to speed up the convergence speed of the agent. Compared with the baseline algorithms in multi-agent pursuit scenarios, our algorithm can converge to the best strategy and has fast convergence speed. In addition, the success rate of the algorithm in partially observable scene is increased by about 20%, and the required time step is reduced by about 25%. Experiments show our algorithm is effective and robust.
  Keywords: reinforcement learning; sparse reward; self-satisfaction; internal emotional reward.
  DOI: 10.1504/IJBIC.2024.10066075
   
  
• Double Fuzzy Clustering Driven Context Neural Network Optimised with Chimp Optimisation Algorithm for Movie Rating Recommendation system  
  by K. Krishnaveni, S. Siva Ranjani 
  Abstract: This paper proposes a pioneering approach called double fuzzy clustering driven context neural network optimised by chimp optimisation algorithm for movie rating recommendation (DFCCNN-COA-MRR). Motivated by the need to enhance recommendation accuracy and mitigate cold start issues, this model integrates double fuzzy clustering with context-aware neural network architecture, bolstered by the chimp optimisation algorithm for weight parameter optimisation. Leveraging the MovieLens 100k dataset, feature extraction and clustering are conducted to form contextual clusters, enabling more precise recommendations. The performance of the proposed DFCCNN-COA-MRR algorithm attains 33.01%, 37.82% and 36.73% high accuracy, 1.16%, 5.07% and 2.71% lower error rate and 32.92%, 35.65% and 33.15% better precision comparing to the existing methods like DRPRA-MRR, EGJSM-CgS-MRR and CAR-RA-MRR respectively. Through this work, contribute a novel recommendation model that successfully addresses key challenges in collaborative filtering, thereby advancing state-of-the-art in recommendation system research.
  Keywords: grapheme-based anisotropic polarisation meta-filter; force-invariant improved feature extraction method; double fuzzy clustering driven context neural networks; chimp optimisation algorithm.
  DOI: 10.1504/IJBIC.2024.10066576
   
  
• Attention Segmental Recurrent Neural Network Optimised with Sheep Flock Optimisation based Intrusion Detection Framework for Securing IoT  
  by M. Ramkumar Raja, P.J. Sathish Kumar, Jayaraj V, Krishnan Somasundaram 
  Abstract: This manuscript proposes an Attention Segmental Recurrent Neural Network (ASRNN) optimized with Sheep Flock Optimization based Intrusion Detection Scheme for securing internet of things (IoT) environment. Initially, the data is fed to preprocessing, wherein, the redundancy eradication and missing value replacements are performed by random forest and local least squares (LLS). Afterward, pre-processing data is supplied to the feature selection to select optimal features. The Correlation feature selection based processing of feature selection is done. The selected features are fed to Attention Segmental Recurrent Neural Network, which categorizes the data as normal or anomalies. Finally, Sheep Flock Optimization (SFO) is considered to optimize the ASRNN. The simulation performance of the proposed technique attains better accuracy 20.56%, 18.67%, 23.77%, 38.45%, 22.75%, 36.45%, higher precision 42.36%, 22.15%, 56.45%, 22.03%, 28.63%, and 21.36% compared with the existing methods.
  Keywords: Wireless networks; Feature selection; Sheep Flock Optimization; Attention Segmental Recurrent Neural Network; Intrusion detection systems.
  DOI: 10.1504/IJBIC.2024.10066580
   
  
• Identification of Primary Central Nervous System Lymphoma from High-Grade Glioma based on a 18F-FDG PET/CT Radiomics Nomogram Compared with Deep Learning: a Multicentre Study  
  by Xin Jin, Yujing Zhou, Lili Qu, Jingtao Wang, Shoumei Yan, Kaiyue Li, Hang Zhou, Li Ma, Xin Li 
  Abstract: Diagnosing central nervous system space-occupying lesions still depend on stereotactic biopsy. Therefore, a non-invasive imaging method is urgently required to distinguish between the two. This retrospective study enrolled 66 patients (38 with PCNSL and 28 with HGG) who underwent 18F-fluorodeoxyglucose-positron emission tomography/computed tomography (18F-FDG PET/CT) between July 2017 and July 2022 to investigate the ability of 18F-FDG PET/CT -based radiomics features in differentially diagnosing PCNSL and HGG. A group of 40 patients was assigned as the training cohort, while another group of 26 patients as the validation cohort. A total of 788 radiomics features were extracted from 18F-FDG PET/CT images in the training cohort. Two features were selected by the LASSO method from 788 features to build the logistic model and radiomics nomogram. The AUC of the radiomics nomogram for discriminating PCNSL from HGG was 0.960 [95% confidence interval (CI): 0.9091] and 0.920 (95% CI: 0.7941) in the training and validation cohorts, respectively. The training and validation revealed that the established radiomics nomogram model of 18F-FDG PET/CT displayed excellent discrimination capabilities in PCNSL and HGG, and may have the potential to improve diagnostic accuracy and patient outcomes.
  Keywords: radiomics; deep learning; 18F-FDG PET/CT; high-grade glioma; HGG; primary central nervous system lymphoma; PCNSL.
  DOI: 10.1504/IJBIC.2024.10066760
   
  
• Multi-Objective Cigarette Production Scheduling Problem  
  by Weidong Lou, Yong Jin, Hailong Lu, Yanghua Gao, Xue Xu 
  Abstract: Cigarette production scheduling in a tobacco enterprise is an optimisation problem that can directly reflect the revenue and production status of the enterprise, so it is necessary to propose a more accurate scheduling model based on the real tobacco enterprise Firstly, this paper proposes a high-dimensional multi-objective cigarette production scheduling model considering time, cost, energy consumption, number of card changes, and load balancing Secondly, a self-adaptive NSGA-III algorithm (SA_NSGA-III) based on indicator guidance is proposed to better solve the modified model and generate a scheduling scheme that can effectively improve the production performance, SA_NSGA-III introduces PD diversity evaluation indicators to guide the population evolution and solve the problem of poor diversity in the late convergence stage of the algorithm Finally, the proposed algorithm is experimentally verified by real data examples from enterprises, and the results show that compared with other comparative algorithms, the SA_NSGA-III algorithm achieves optimal results.
  Keywords: cigarette production; scheduling; optimisation; evolutionary algorithm.
  DOI: 10.1504/IJBIC.2024.10066764
   
  
• Synthesis of Unequally Spaced Linear Antenna Array for Subsidiary Maxima Elimination Using LGChOA Algorithm  
  by Simrandeep Singh, Harbinder Singh, Amit Gupta, Ahmed Jamal Abdullah Al-Gburi 
  Abstract: The chimp optimisation algorithm (ChOA) draws inspiration from the individual intellect of chimps during their collective hunting, setting them apart from other social predators. Despite its potential, ChOA often encounters premature convergence to local optima during the search phase, limiting its effectiveness in balancing exploitation and exploration and remains susceptible to stagnation. In this research, an enhanced modified version of ChOA is proposed, integrating Levy flight and greedy selection procedures to address these shortcomings. The efficacy of the proposed algorithm is evaluated in the context of antenna array synthesis. Linear antenna arrays find extensive use in wireless communication applications, yet achieving the dual objectives of suppressing subsidiary maxima while maintaining sufficient spacing and side lobes poses a significant challenge. The proposed approach is evaluated across various linear array communication requirements, and the results are compared with those obtained using other well-known strategies.
  Keywords: chimp optimisation algorithm; grating lobe; linear antenna arrays; side lobe level; subsidiary maxima.
  DOI: 10.1504/IJBIC.2024.10066765
   
  
• Segmentation and Classification of Brain Tumour with Optimisation Enabled Deep Learning Using MRI Images  
  by Sajeev Ram Arumugam, Sakthi Ulaganathan, Rajeshkannan Regunathan, Vimala S. 
  Abstract: The detection of brain tumour at final stage is difficult to heal and the diagnosis of brain tumour from large image database is difficult. Due to the various sizes, shapes, and locations of tumours in the brain, the present techniques are insufficient to give precise classification. Hence, an effective model for classifying and segmenting brain tumours is developed in this study using the circle inspired teaching learning optimisation method (CITLO). During the first step, an input MRI image from a dataset is obtained, and the obtained image is supplied into the pre-processing module. Following that, SegNet, which is trained using CITLO, is employed for tumour segmentation. The brain tumour classification process employs deep convolutional neural network (DCNN), with classifier hyper parameters learned using CITLO. The CITLO_DCNN attained a maximum accuracy of 95.8%, a sensitivity of 96.9%, a specificity of 96.6%, a maximum segmentation accuracy of 95.7%, and ROC of 93.1%.
  Keywords: SegNet; MRI image; circle inspired teacher learning optimisation; deep learning; tumour segmentation; deep convolutional neural network; DCNN.
  DOI: 10.1504/IJBIC.2024.10066793
   
  
• A Chaotic Simulated Annealing Genetic Algorithm with Asymmetric Time for Offshore Wind Farm Inspection Path Planning  
  by Lei Kou, Yukuan Wang, Fangfang Zhang, Quande Yuan, Zhen Wang, Jingya Wen, Wende Ke 
  Abstract: The harsh environment of offshore wind farms causes wind turbines to be easily broken down. To ensure the normal operation of wind turbines, it is necessary to carry out inspections of offshore wind farms. Path planning is an important step to complete the inspection. The unmanned surface vessel (USV) is subject to the water current, leading to deceleration and acceleration, which makes the round-trip travelling time of the USV between two wind turbines asymmetric. To sum up, it belongs to asymmetric travelling salesman problem. To address this problem, a chaotic simulated annealing genetic algorithm (CSAGA) considering asymmetric time is proposed in this paper. Firstly, the initial sequence with high quality, as the initial solution of the simulated annealing algorithm, is generated by logistic-tent chaotic mapping. Then, effective solutions are obtained by a series of operations of the simulating annealing algorithm and is used to replace the worst fitness individuals in the initial population of the genetic algorithm. Finally, genetic operations such as selection, crossover, mutation, and reversion are applied to the population to obtain the optimal solution. The feasibility of the algorithm is verified by simulation experiments. The results display that CSAGA has better performances compared to other algorithms.
  Keywords: travelling salesman problem; TSP; inspection path planning; simulated annealing algorithm; genetic algorithm; offshore wind farm.
  DOI: 10.1504/IJBIC.2024.10066897
   
  
• Multi-Objective Jaya War Strategy Optimisation Enabled Cloud Storage in Blockchain Network for Internet of Medical Things Applications  
  by Kavita Shelke, Subhash Shinde 
  Abstract: The internet of medical things (IoMT) includes accessibility and is adequate for medical areas. Diverse techniques are developed based on cloud storage for IoMT applications, but these methods failed to obtain sufficient security in less time. In this research, a Jaya war strategy optimisation (Jaya WSO)-based cloud storage in a blockchain network is introduced for IoMT. Firstly, transactions are generated in IoMT and fed to the base station (BS). Then, it is passed to peers in the blockchain and the ledger is stored in respective peers. For each peer, blocks are selected optimally by Jaya WSO, which is an integration of the Jaya algorithm with war strategy optimisation (WSO) and the multi-objectives. The Jaya WSO attained minimal query probability, storage cost and local space occupancy of 0.428, 20.766 and 52.7 respectively and maximal values of trust level and sensitivity level of 0.849 and 0.916 respectively for block size = 2.
  Keywords: internet of medical things; IoMT; blockchain; Jaya algorithm; war strategy optimisation; WSO; cloud storage.
  DOI: 10.1504/IJBIC.2024.10066976
   
  
• Natural Nearest and Shared Nearest Neighbors Density Peaks Clustering Algorithm for Manifold Data  
  by Lu Li, Zhigang Li, Shenyu Qiu, Zhaoxiu Nie, Han Longzhe 
  Abstract: Addressing the challenges faced by the density peaks clustering (DPC) algorithm in precisely identifying cluster centres when dealing with manifold data and its tendency to misclassify samples distant from cluster centres, this paper proposes a natural nearest and shared nearest neighbours' density peaks clustering algorithm for manifold data. The local density is redefined by natural nearest and shared nearest neighbours, which highlights the difference between the cluster centre and other samples, and makes the identification of the cluster centre more accurate; the similarity matrix is constructed according to the similarity between the samples defined in the process of defining the local density to complete the allocation of the remaining samples, which effectively improves the problem of incorrectly allocating samples far away from the centre of the clusters in the manifold clusters. Upon comparing the algorithm described in this paper with DPC and other refined algorithms, the experimental outcomes stemming from the manifold datasets unambiguously demonstrate its proficiency in accurately pinpointing the centroid of each cluster, thus effectively carrying out the clustering task. At the same time, on the UCI datasets and Coil20 dataset, this paper's algorithm can get an ideal clustering effect.
  Keywords: density peaks clustering; DPC; manifold data; natural nearest neighbours; NNN; shared nearest neighbours; SNN.
  DOI: 10.1504/IJBIC.2024.10067699
   
  
• Solving Bank Debt Problems based on Parallel NSGA-II Algorithm  
  by Xuezhi Yue, Teng Xiong, Wenxing Zhu 
  Abstract: In order to alleviate the balance of bank liquidity, profitability, and safety, this paper regards bank liability management as a multi-objective optimisation problem, establishes a multi-objective bank liability model with solvency, liquidity risk, and net interest income as the goals, and proposes an improved adaptation value method and environment selection method to improve the NSGA-II algorithm (PNSGA-II) to realise the optimisation of bank asset management. Compared with the NSGA-II algorithm, the PNSGA-II algorithm has better convergence and diversity, so as to better solve the problem of bank liability management. Compared with the NSGA-II algorithm, SPEA2 algorithm, and improved algorithm, the Pareto frontier distribution and IGD index of the PNSGA-II algorithm have better performance, indicating that the proposed algorithm has better convergence and diversity, and better comprehensive performance. The experimental results show that by using the parallel NSGA-II algorithm to solve the bank liability problem, banks can select a realistic set of optimal solutions according to the actual situation among the six sets of Pareto optimal solutions, so as to more conveniently and objectively predict the liability management, asset allocation, and macro-control in the next few years.
  Keywords: asset liability management; multi-objective optimisation; PNSGA-II; prioritisation.
  DOI: 10.1504/IJBIC.2024.10067776
   
  
• Design of Logic Circuit for All Optical Sampling Gate Based on FIBER-FWM Nonlinear Fiber  
  by Yunhu Wu, Yarang Yang, Wei Zheng, Saidiwaerdi Maimaiti, Hui Peng 
  Abstract: To solve the problems of limited sampling bandwidth and low sampling rate in existing optical sampling techniques, an all optical sampling gate logic circuit design based on FIBRE-FHM nonlinear fibre is adopted to measure high-speed optical signals, improving the bandwidth of traditional sampling methods. This design addresses significant limitations posed by electronic bottlenecks in high-speed optical signal measurement. Utilising the four-wave mixing (FWM) model of a semiconductor optical amplifier (SOA), a sampling gate is realised via a logic circuit leveraging nonlinear effects. These results confirm that after nonlinear processing, the pulse width increases by 80% to 0.1
  Keywords: light sampling; nonlinear fibre; logic circuit; semiconductor optical amplifier; SOA; four-wave mixing; FWM.
  DOI: 10.1504/IJBIC.2024.10067779
   
  
• Attention Mechanism-Based Facial Age Estimation  
  by Zhang Huiying, Lin Jiayan, Sheng WenShun, Dong Jiangwei, Zhang Yu, Geng Xin, Deyin Zhang, Jin Xin 
  Abstract: With the development of the deep learning (DL) technique, especially Long Short-Term Memory (LSTM) for personal aging patterns, the accuracy of facial age estimation has been significantly improved. However, in traditional DL framework, the interdependence between individual facial images has not been fully exploited. To improve the estimation accuracy further, we propose an attention mechanism-based face aging estimation (AM-FAE) to characterize such meaningful interdependence. The proposed AM-FAE is able to select the most relevant parts of the input and assigns different weights to different contextual face information, thereby can achieve high value of information. Compared to state-of-the-art facial age estimation methods, AM-FAE improves the accuracy of age estimation on two public datasets.
  Keywords: mechanism of attention; Convolutional Neural Networks; Label Distribution Learning; facial age estimation.
  DOI: 10.1504/IJBIC.2024.10067926