International Journal of Complexity in Applied Science and Technology (23 papers in press)

Regular Issues

• Hybrid CNN with Chebyshev Polynomial Expansion for Medical Image Analysis  
  by Abhinav Roy, Bhavesh Gyanchandani, Aditya Oza 
  Abstract: Lung cancer remains a leading cause of cancer-related deaths, where early and accurate diagnosis is vital. Automated detection of pulmonary nodules in CT scans is challenging due to variations in nodule characteristics. While CNNs have shown promise, they struggle to capture fine-grained spatial-spectral features. We propose a hybrid CNN architecture enhanced with Chebyshev polynomial expansions, leveraging their orthogonality and approximation properties to extract high-frequency features and improve non-linear function modeling. Evaluated on LUNA16 and LIDC-IDRI datasets, our model outperforms standard CNNs in classifying nodules as benign or malignant, achieving notable gains in accuracy, sensitivity, and specificity. This method offers a robust framework for medical image analysis and clinical decision support.
  Keywords: Chebyshev Polynomials; Convolutional Neural Networks; Deep Learning; Function Approximation; Hybrid Deep Learning Model.
  DOI: 10.1504/IJCAST.2026.10074363
   
  
• A Comparative Study of Machine Learning and Grey Approach for Forecasting FX Rates  
  by Noorshanaaz Khodabaccus, Aslam A. E. F. Saib 
  Abstract: Volatility in the foreign exchange (FX) market is often associated with risk and can disrupt the sustainable development of an economy. The Mauritian economy, being open and globally integrated, is highly sensitive to currency fluctuations. Consequently, modelling and forecasting FX market volatility is crucial for proper risk management. This paper presents a comparative study on FX rate modelling and forecasting accuracy, contrasting conventional deep learning approaches with grey models. In particular, we compare the performance of recurrent neural networks (RNNs) and long short-term memory (LSTM) network,s using high-frequency historical data against the basic grey model and the optimised Fourier grey Markov model (FOGM), which relies on a significantly smaller dataset. Our findings indicate that the FOGM model, despite using a smaller dataset, outperforms the deep learning approaches considered.
  Keywords: FX rates forecasting; Volatility; Deep learning; Grey model; Optimised Fourier grey Markov model.
  DOI: 10.1504/IJCAST.2025.10074547
   
  
• Advancing Deep Learning Techniques for Low-Resource Shahmukhi Punjabi Language Processing  
  by Muhammad Shabbir, Mudassir Iftikhar 
  Abstract: Shahmukhi Punjabi, over 18 million of Pakistani speaks the Shahmukhi Punjabi language but there is not proper research is being available yet. So this paper explores the software of named entity recognition (NER), recurrent neural network (RNN), and long short-term memory (LSTM) models on a dedicated dataset. The study includes a thorough analysis of loss graphs, accuracy measures, and understanding matrices. Our contribution is looks into Shahmukhi Punjabis underappreciated work on a variety of sophisticated complex space models, including NER and network RNN. The majority of current research focuses on languages that are widely spoken. By implementing LSTM, RNN, and NER models and evaluating their efficiency on specific Shahmukhi Punjabi data, this work seeks to close this gap on Shahmukhi language research. Our LSTM models give us the 82% accuracy and RNN give us the 82.57%.
  Keywords: NLP; Deep Learning; RNN; LSTM; Data Science.
  DOI: 10.1504/IJCAST.2026.10075028
   
  
• A Model-Controller-Presenter Tri-Layer Control-Theoretic Orchestration Framework for Synergistic Efficiency and Interpretability in Multimodal Large Language Models  
  by Yaolin Zhang, Menghui Li 
  Abstract: Targeting the issues of insufficient computational efficiency and limited interpretability encountered by large-scale models when dealing with complex tasks such as multi-turn reasoning and multi-modal cooperation, this research puts forward a three-tier collaborative framework centred on model-controller-task adaptation (MCP). By decoupling the functionalities of large models into reasoning, generation, and retrieval modules, and integrating dynamic routing algorithms driven by reinforcement learning and task adaptation mechanisms, the systematic integration of control theory and the dynamic reasoning of large models is realised for the first time. Experimental results demonstrate that the MCP framework enhances the performance on cross-modal benchmark tasks (e.g., GLUE, COCO, ScienceQA) by 1530% compared with the baseline model, improves reasoning efficiency by 40%, and generates interpretable intermediate results through the presenter layer, achieving a manual interpretability score of 90%. This offers a brand-new technological route to tackle the bottleneck in the practical application of large-scale models.
  Keywords: large model optimisation; MCP framework; dynamic control flow; reinforcement learning routing;.
  DOI: 10.1504/IJCAST.2025.10075441
   
  
• AI-Driven Prediction Model for Antenatal and Postpartum Depression Among Bangladeshi Pregnant Mothers  
  by M.D. Zahurul Haque, Tasnim Binta Anowar, Sumaiya Jannat Samira 
  Abstract: Antenatal and postpartum depression (APD) are significant maternal mental health concerns, especially in lowresource settings like Bangladesh. This study introduces an AI-driven prediction model aimed at identifying the risk of APD among Bangladeshi mothers. Data were gathered from over 500 participants via Google Forms distributed through hospitals, online platforms, and community networks. The dataset encompasses a range of demographic, psychological, and lifestyle factors. Machine learning algorithms-random forest, XGBoost and gradient boosting were employed, demonstrating high accuracy in predicting depression severity. The developed web-based application enables real-time risk assessments, facilitating early detection and timely intervention. This research highlights the transformative role of AI in enhancing maternal mental health services and delivering scalable, data-driven solutions in resource-limited environments.
  Keywords: Antenatal depression; Postpartum depression; Machine learning; Mental health prediction; AI in healthcare.
  DOI: 10.1504/IJCAST.2025.10075720
   
  
• Comparative Study between a GA-Tuned PID Controller based on Minimising IAE, ITAE, and ISE Objective Functions in a Shell and Tube Heat Exchanger Temperature Control System  
  by Melat Getachew Kebede  
  Abstract: A heat exchanger device is widely used in process industries because it is capable of sustaining a wide range of temperatures. The heat exchanger temperature control system is a highly nonlinear, time-delayed, and complex system. additionally, it is accompanied by the presence of flow variation and temperature variation of input fluid. In this study, performance of a genetic algorithm-based proportional integral and derivative controller tuned based on minimising IAE, ITAE, and ISE objective functions has been analysed and compared under four scenarios, namely no disturbance, flow variation disturbance, temperature variation disturbance, and both disturbance conditions. Among the three objective functions used, the overall performance of the genetic algorithm-based proportional integral and derivative controller tuned based on minimising the time integral of the absolute error objective function is better than the one tuned based on minimizing the integral of the absolute error and the integral of the squared error objective functions.
  Keywords: : Genetic algorithm; GA tuned PID controller; Integral Square Error; Integral Absolute Error and Integral Time Absolute Error.
  DOI: 10.1504/IJCAST.2025.10076038
   
  
• A Comprehensive Analysis of Classical Sorting Algorithms with Diverse Input Conditions  
  by Mohsen Mohammadagha 
  Abstract: This study presents a comprehensive analysis of six classical sorting algorithms Mergesort, Heapsort, Quicksort (with median-of-three pivot), Insertion Sort, Selection Sort, and Bubble Sort to evaluate their practical efficiency across diverse input conditions. While theoretical complexity (O(n2) vs. O(nlog(n))) provides foundational insights, real-world performance depends on implementation-specific factors, input size, and data distribution. The research addresses the critical need to bridge theoretical predictions with empirical benchmarks, particularly as modern computing environments demand optimised algorithm selection for varying workloads. Using Python-based implementations, the methodology systematically tests algorithms on arrays (size 10100,000) with randomised, sorted, reverse-sorted, and custom patterns, measuring execution times and memory usage. Results reveal quadratic algorithms outperform O(nlog(n)) methods for small datasets (e.g., Selection Sort: 0.000004s at n = 10), while Quicksort dominates at scale (0.089s vs. Bubble Sorts 265.93s at n = 100,000). Logarithmic visualisations highlight exponential efficiency divergence, with O(nlog(n)) algorithms achieving 2,900
  Keywords: Modeling; Optimization; Hybridization; Sorting Algorithms; Comparative Study.
  DOI: 10.1504/IJCAST.2025.10076087
   
  
• Quantum AI-Driven Cloud Framework for Intelligent Urban Surveillance  
  by Ranjan Kumar Mandal  
  Abstract: Rapid urbanization has intensified challenges in public safety, efficient resource allocation, and effective governance. Smart city initiatives address these concerns through advanced technological frameworks, with cloud computing and artificial intelligence (AI) playing critical roles in modern surveillance solutions. However, existing systems face limitations in real-time processing and timely decision-making when handling large-scale data streams. This paper presents a Quantum AI-Integrated Cloud Surveillance Architecture designed for smart cities. The proposed framework leverages quantum-enhanced computational capabilities within cloud infrastructure to accelerate large-scale data processing and improve scalability. Quantum AI algorithms further enhance analytical performance, enabling real-time video analysis, anomaly detection, and predictive modelling. Experimental evaluations demonstrate the system's potential to deliver proactive threat mitigation and optimised resource management, thereby offering a robust foundation for next-generation urban surveillance.
  Keywords: Smart city; cloud computing; quantum computing; artificial intelligence; surveillance systems; real-time analytics; anomaly detection; predictive modelling.
  DOI: 10.1504/IJCAST.2026.10076619
   
  
• A Comparative Study of NLP Systems for Sentiment Polarity Classification across Different Domains and Genres  
  by Vincenzo Sammartino 
  Abstract: Sentiment analysis systems are now among the most widely used tools across various sectors: from politics to stock markets, from marketing to communication, from the sports domain to medical and natural sciences, and from social media analysis to consumer preference evaluation. This study presents a performance comparison of different methodologies, techniques, and applications developed in recent years. We describe a concrete implementation of two distinct Natural Language Processing (NLP) systems for the sentiment polarity classification of Italian tweets and Amazon reviews. Two different classification systems were developed: the first employs an approach based on the explicit representation of the texts' linguistic features, while the second uses an approach based on non-interpretable vectors (embeddings). Finally, a study was conducted to understand which features are most relevant for classification, and the underlying causes that influence the systems' performance in both in-domain and out-of-domain scenarios are highlighted.
  Keywords: Sentiment Analysis; Natural Language Processing; Machine Learning; Support Vector Machines; Domain Adaptation; Text Classification; Bag-of-Words; Word Embeddings; Feature Engineering; Italian; UGC.
  DOI: 10.1504/IJCAST.2026.10076620
   
  
• Feature Engineering and Machine Learning Framework for DDoS Attack Detection in the Standardised Internet of Things  
  by N.S. Akash, V. G. Naveen Kumar 
  Abstract: Distributed Denial of Service (DDoS) attacks are a major threat to cloud servers, and the rapid growth of Internet of Things (IoT) devices has further intensified this problem. Large-scale IoT-based DDoS attacks can overwhelm networks and disrupt essential services. To address this, we present a machine learning driven, multi-layer detection framework that integrates IoT devices, Gateways, software-defined networking (SDN) switches, and cloud servers. For experimentation, we deployed eight smart poles on our campus equipped with diverse sensors and gathered real-time data through both wired and wireless networks. Features relevant to different categories DDoS attacks were extracted and used to train machine learning models, achieving high detection accuracy in realistic IoT environments. Our results demonstrate that the proposed framework can effectively identify malicious traffic and leverage SDN controllers to block compromised devices in real-time.
  Keywords: Distributed-Denial of Service? Internet of Things; Machine Learning; Software Defined Networking.
  DOI: 10.1504/IJCAST.2025.10077283
   
  
• Learning Socially-Aware Navigation via Predicate-Grounded Perception and Intent Modelling  
  by W.U. JiaHao 
  Abstract: Navigating among humans requires robots to understand and respect social conventions around personal space, movement patterns, and implicit intentions. We address this challenge by learning symbolic predicates that ground multi-modal perceptual data into interpretable representations suitable for social reasoning. Our approach trains a neural network to predict spatial relationships (proximity, velocity, risk) from LiDAR, vision, and depth sensors, then applies structured rules to infer human intent and generate socially-compliant actions. Through experiments in hospital and commercial environments with both static and dynamic obstacles, we demonstrate that predicate-grounded perception substantially outperforms end-to-end learning and classical planning methods in safety, efficiency, and interpretability. Analysis of failure modes reveals challenges in crowded scenarios and limitations of hand-crafted reasoning rules, suggesting directions for future work in learned symbolic reasoning.
  Keywords: Social navigation; symbolic grounding; human--robot interaction; intent recognition; multi-modal perception; explainable AI; neuro-symbolic learning; robot motion planning.
  DOI: 10.1504/IJCAST.2026.10077874
   
  
• Dynamic Rank Reinforcement Learning for Adaptive Low-Rank Multi-Head Self-Attention in Large Language Models  
  by Caner Erden 
  Abstract: Dynamic rank reinforcement learning (DR-RL) approximations rely on static rank assumptions, limiting their flexibility across diverse linguistic contexts. Our method dynamically modulates ranks based on real-time sequence dynamics, layer-specific sensitivities, and hardware constraints. The core innovation is a deep reinforcement learning agent that formulates rank selection as a sequential policy optimisation problem, strictly balancing attention fidelity against computational latency. To ensure stability during inference, we derive and employ online matrix perturbation bounds, enabling incremental rank updates without the prohibitive cost of full decomposition. Furthermore, the integration of a lightweight transformer-based policy network and batched singular value decomposition (SVD) operations ensures scalable deployment on modern architectures. Extensive experiments demonstrate that DR-RL significantly reduces floating point operations (FLOPs) by over 40% in long-sequence regimes (L > 4,096) while maintaining downstream accuracy statistically equivalent to full-rank attention. Beyond standard language modelling benchmarks, we validate the real-world applicability of DR-RL on the GLUE benchmark. Specifically, our method achieves 92.78% accuracy on the SST-2 sentiment analysis task, matching the performance of full-rank baselines and outperforming static low-rank methods, such as Performer and Nystr
  Keywords: Large Language Models; LLM; Multi-Head Self-Attention; Reinforcement Learning; Low-Rank Approximation; Dynamic Rank Selection.
  DOI: 10.1504/IJCAST.2026.10077946
   
  
• Genetic algorithm-based optimal placement of reclosers in Chiro power distribution  
  by Melat Getachew Kebede, Umer Abdo Geru 
  Abstract: The reliability of an electrical power system indicates its capability to supply stable and continuous electricity that satisfies consumer needs. This study evaluates the reliability performance of the Chiro substation (Hirna feeder) using data obtained from the local utility. The calculated reliability indices show a system average interruption frequency index (SAIFI) of 255.03 interruptions per customer and a system average interruption duration index (SAIDI) of 257.58 hours per customer. The customer average interruption duration index (CAIDI) is 1.010 hours, with an average service availability index of 0.9706 Pu and an unavailability index of 0.0294 Pu. The energy not supplied (ENS) is 711.577 MWh per year, signifying poor reliability. To enhance system performance, the study applies a genetic algorithm (GA) optimisation in MATLAB to determine the optimal placement of reclosers, considering interruption frequency as the objective. The optimal configuration consists of ten reclosers positioned at specific distances along the feeder.
  Keywords: genetic algorithm; reclosers; optimal placement; power distribution system; reliability indices.
  DOI: 10.1504/IJCAST.2026.10078180
   
  
• Multilingual Fake News Detection in Indic Languages Using Fine-Tuned MuRIL  
  by A.V.I. Verma, Abhinandan Rathi, Aditya Verma 
  Abstract: The rapid spread of fake news in low-resource Indic languages poses a significant challenge to information integrity on social media platforms in India. This paper presents a multilingual fake news detection system tailored for Indic languages, leveraging the pre-trained multilingual representations for Indian languages (MuRIL) transformer model. We fine-tune MuRIL on a large-scale dataset comprising 82,946 news statements across multiple Indic scripts, including English, Hindi, Tamil, Gujarati, Malayalam, Punjabi, Bengali, Telugu, Marathi, and others. Experimental results on a stratified 8020 train-test split demonstrate strong performance, achieving an accuracy of 83.86%, precision of 85.36%, recall of 90.62%, and F1-score of 87.91%. A detailed language-wise analysis reveals robust performance on high-resource languages like English while highlighting limitations in low-resource ones due to data imbalance. The model is deployed as a real-time Gradio application on Hugging Face Spaces for public use. Our contributions include a comprehensive multilingual dataset analysis, fine- tuned MuRIL model, and in-depth language-specific error diagnosis.
  Keywords: Fake News Detection; MuRIL; Indic Languages; Multilingual NLP;Transformers; Transformer Models; Deep Learning; Social Media; Misinformation;Code Mixing.
  DOI: 10.1504/IJCAST.2026.10078584
   
  
• A Review of Artificial Intelligence Transforming Education through: Technologies, Applications, and Transformative Impact  
  by Khazar Gorji, Reza Saeed Kandezy 
  Abstract: convolutional neural networks, recurrent neural networks, transformer models, generative adversarial networks, and reinforcement learning and evaluates their applications across STEM, language learning, art education, and lifelong learning. We examine domain-adaptations, scalability issues, interpretability, fairness, and ethical deployment, and propose design principles in the spirit of cognitive-affective-social learning theories. Cross-modal data fusion, hybrid neuro-symbolic and teacher-interactive explainability methods are identified as key enablers for secure AI integration. To bridge the gap between algorithmic innovation and the real-world classroom effect, we recommend a multidimensional assessment framework that considers predictive reliability, pedagogical value, ethical compliance, and deployability. We also emphasise the importance of standardised metrics and longitudinal studies. Ultimately, we argue that AIs transformative power in schools relies as much on pedagogical alignment, cultural sensitivity, and human-AI collaboration as it does on technological advancements.
  Keywords: Artificial Intelligence; Education; STEM; Cognitive-Affective Learning; Machine Learning; Convolutional Neural Networks.
  DOI: 10.1504/IJCAST.2026.10078654
   
  
• Simulation of a standalone photovoltaic street lighting system using ANFIS based maximum power point tracking strategy  
  by Melat Getachew Kebede, Umer Abdo Geru 
  Abstract: The increasing demand for electrical energy and the limitation of traditional energy sources have increased the significance of renewable energy sources. In Ethiopia there are frequent power outages, especially during peak times. The direct solution is to cut some of its load by designing a stand-alone street light. The main objective of this study is to simulate a stand-alone photovoltaic street lighting system using adaptive neuro-fuzzy inference system-based maximum power point tracking. All the components of the system are modeled in MATLAB/Simulink. The irradiance data used in this study are taken from Mieso and Asebot towns. The simulation result depicts improved tracking of the maximum power and stable battery operation. The finding of this research indicated that the ANFIS controller improves the efficiency and sustainability, offering a promising solution for grid-disconnected street lighting in rural and semi-urban areas of Ethiopia.
  Keywords: Street Lighting; Photovoltaic; ANFIS Controller; Renewable Energy Sources.
  DOI: 10.1504/IJCAST.2026.10078665
   
  
• Financial Market Volatility Using the SVR-GARCH for Frontier Market Equities  
  by Carl Hope Korkpoe  
  Abstract: Financial market volatility is a critical factor influencing investment decisions, risk management, and economic policy. Traditional statistical models often struggle to capture the complex, nonlinear relationships in financial data, leading to suboptimal volatility forecasting. Machine learning (ML) techniques presently offer a promising alternative by leveraging vast amounts of historical data to identify hidden patterns and improve predictive accuracy. We explore the use of a hybrid Support Vector Regression (SVM)-GARCH algorithm against the various variant of GARCH to study the heteroscedasticity of the equity returns drawn from the Ghana Stock Exchange (GSE) as typically representing frontier markets in sub-Saharan African markets. We compared the effectiveness of these hybrid models to the classical GARCH and highlighted the advantages of data-driven ML models in adapting to dynamic market conditions. Our findings demonstrate that ML-based models can enhance forecasting performance, reduce estimation errors, and provide deeper insights into market behavior, making them valuable tools for both investors and policymakers in frontier and developed markets alike.
  Keywords: Machine learning; Support Vector Machine; Financial market volatility; GARCH; Frontier markets.
  DOI: 10.1504/IJCAST.2026.10078685
   
  
• Multi-Class Sentiment and Toxicity Analysis of YouTube Comments Using Hybrid Machine Learning and Transformer Models  
  by A.V.I. Verma 
  Abstract: The rapid growth of user-generated text on digital platforms requires scalable and real-time systems for sentiment analysis and toxicity detection. Existing approaches often rely on coarse sentiment labels or standalone deep models that struggle with noisy short text and deployment challenges. This paper presents a hybrid framework that combines classical machine learning with transformer-based models. A Random Forest classifier is trained on 28 fine-grained emotion categories and integrated with Twitter-RoBERTa for polarity classification and Toxic-BERT for toxicity detection. The pipeline includes batch inference optimization, fault tolerance with retries and rate limiting, Redis caching, VADER-based fallback for low-confidence predictions, and Cohen’s Kappa agreement monitoring. Experimental results show the hybrid model achieves 87% accuracy and 0.86 macro F1-score, outperforming baseline models while maintaining low latency. Deployed as a Gradio application on Hugging Face Spaces with YouTube API integration, the system supports large-scale real-time comment analysis.
  Keywords: Sentiment Analysis; Hate Speech Detection; YouTube Comments; Random Forest; Transformer Models; Hy brid Machine Learning; Gradio Deployment; Toxicity Detection; Natural Language Processing.
  DOI: 10.1504/IJCAST.2026.10079338
   
  
• Efficient Self-Verification for Tool-Augmented Language Models via Consistency-Based Reward Modelling  
  by Abdulloh Nidhom 
  Abstract: Tool-augmented large language models can function as autonomous research agents, but their reliability is often bottlenecked by costly verification pipelines that depend on large judge models or human annotation. Reinforcement learning from AI feedback typically requires heavy reward models, putting it out of reach for many researchers. ConsistencyVerify offers a lighter alternative: it uses multi-sample agreement and uncertainty estimates to generate stable reward signals without external judges. By relying on temperature-scaled sampling and agreement-based scoring, the method captures the model’s own calibration structure. Agents trained with ConsistencyVerify match the performance of judge-based approaches on Natural Questions, TriviaQA, and HotpotQA while running on a single consumer GPU. Ablation studies indicate that consistency-based signals produce smoother training than lexical metrics such as F1 or ROUGE. This suggests a practical route toward more accessible and computationally efficient research agent training.
  Keywords: tool-augmented LLMs; self-verification; consistency-based reward modeling; RLAIF; uncertainty estimation; multi-sample agreement; semantic similarity; research agents; efficient training.
  DOI: 10.1504/IJCAST.2026.10079341
   
  
• Decomposable Reward Modeling and Realistic Environment Design for Reinforcement Learning-Based Forex Trading  
  by Nabeel Saidd 
  Abstract: Reinforcement learning (RL) for foreign exchange trading requires realistic execution modelling, interpretable reward design, and market-aware action spaces. This paper presents a modular RL framework for Forex trading that integrates three components: a friction-aware execution engine with strict anti-lookahead semantics and realistic transaction costs an interpretable 11-component reward architecture with fixed weights and per-step logging for attribution and ablation analysis and a ten-action discrete interface with legal-action masking to enforce margin-aware trading constraints. Experiments on EURUSD training data show complex interactions among reward components, where additional penalties do not always improve performance. The full configuration achieves a training Sharpe ratio of 0.765 and a cumulative return of 57.09%. Extended action spaces increase returns but reduce Sharpe ratios relative to conservative alternatives. Additional tests on GBPUSD, USDJPY, and AUDUSD show similar trends, highlighting the importance of interpretable and execution-realistic RL design in financial systems.
  Keywords: reinforcement learning; foreign exchange trading; reward decomposition; action masking; pyramiding; martingale scaling.
  DOI: 10.1504/IJCAST.2026.10079408
   
  
• Machine Learning Driven Catalyst Selection: Identifying Nickel's Advantage Over Palladium for Suzuki Miyaura Cross Coupling  
  by Kushal Raj Roy  
  Abstract: The Suzuki-Miyaura cross-coupling reaction remains one of the most widely used C-C bond-forming transformations in synthetic chemistry. Machine learning models promise to accelerate reaction optimization, yet systematic benchmarking across different catalytic systems remains limited. Here, we develop and validate a comprehensive machine learning framework for predicting reaction yields across five metal catalysts (Pd, Ni, Ru, Fe, Cu) using a dataset of 5,760 reactions modeled on high-throughput experimentation platforms. Our XGBoost model achieves R2 = 0.903 (RMSE = 6.10%), substantially outperforming transformer-based YieldBERT (R2 = 0.81, RMSE = 11%) while approaching graph neural network performance with dramatically lower computational costs. Systematic catalyst comparison reveals that nickel catalysis achieves superior performance (46.7% mean yield, 42% success rate) compared to conventional palladium (45.8% mean yield, 40% success rate), particularly for challenging chloride electrophiles.
  Keywords: Suzuki-Miyaura coupling; machine learning; yield prediction; nickel catalysis; reaction optimization; cross-coupling; XGBoost; high-throughput experimentation; palladium alternatives;.
  DOI: 10.1504/IJCAST.2026.10079563
   
  
• ReactionForge: A Temporal Graph Network for Suzuki-Miyaura Cross-Coupling Yield Prediction  
  by Kushal Raj Roy  
  Abstract: Accurate reaction yield prediction is critical for synthesis optimization, yet existing models struggle with temporal dynamics, uncertainty quantification, and reactant-product transformation modeling. We present ReactionForge, a Temporal Graph Network integrating five innovations: gated recurrent unit-based temporal memory for catalyst and reagent dynamics, cross-attention layers comparing reactant and product molecular graphs, hierarchical graph pooling for functional group discovery, evidential deep learning for calibrated uncertainty, and multi-task learning across yield, selectivity, and reaction time. Evaluated on 5,760 Suzuki-Miyaura reactions across five metal catalysts, ReactionForge achieves R2= 0.968
  Keywords: Temporal graph networks; reaction yield prediction; Suzuki-Miyaura coupling; deep learning; catalyst optimization.
  DOI: 10.1504/IJCAST.2026.10079754
   
  
• Forecasting the Species of Major Contagious Viruses Through Deep Learning and Data Wrangling Methodology  
  by Vaishnavi Poti, Ashmit Srivastava, Dhruv Tater, Madhuri Rao 
  Abstract: In the context of viral outbreaks, timely identification of viral species plays a crucial role in aiding researchers and healthcare professionals to accelerate vaccine development and formulate treatments. This paper presents a hybrid approach for accurate viral species classification based on genomic sequences. We have made use of a combination of convolutional neural networks (CNN) and long short-term memory (LSTM) networks in our model to extract both local patterns and global dependencies from viral genomic data. We have utilised bioinformatics software and data wrangling methods to build a comprehensive dataset from the National Center for Biotechnology Information and GenBank database. The model achieved an accuracy of 98.46% in classifying viral species. This study provides a computationally efficient approach for the analysis of viral genomic data and provides insight into the prediction of viral evolution. It has the potential to assist in quick response actions in the event of disease outbreaks.
  Keywords: Virus Classification; Deep Learning; Genomic Data; Bio informatics; Machine Learning.
  DOI: 10.1504/IJCAST.2026.10080014