Technology Business Incubation (TBIs) has become a global phenomenon integral to the growth of regionalinnovation and startup ecosystems. The availability of high-quality infrastructure and facilities lays the foundations of theentire startup ecosystem for providing essential support services that directly impact entrepreneurial success. Theincubation capacity of TBIs across different regions can foster competition and collaboration among these regions,provide avenues for enhancing enterprises’ incubation capabilities, and assist entrepreneurs in assessing the strength ofregional incubation. However, with their rapid expansion, the performance evaluation also becomes increasingly complexdue to the diversity of converging factors such as complex technologies, varying nature of relationships of VCs, andentrepreneurial competencies of the founders incubating startups at the TBIs. Traditional Machine Learning performanceevaluation and prediction models struggle to capture these dynamic variables, while also suffering from privacyvulnerabilities, low accuracy, and reliance on centralized third parties. This often leads to single points of failure,performance bottlenecks, and sometimes increased costs. To address these challenges, we employed Privacy-PreservingFederated Learning with Blockchain (PPFL-BC), a novel framework designed for improving the mechanism ofperformance measurement and prediction for remote TBIs while ensuring that the privacy of entities and the dataremains secure. We utilize capabilities of Artificial Neural Network (ANN) and gradient boosting-enabled federatedlearning to train the model of each TBI locally. In the process, no private and sensitive business data is shared outside thenetwork, significantly reducing the risk of privacy breaches. Besides this, all the locally trained models are aggregatedinto a unified predictive model at the central aggregation unit, which ultimately improves the overall accuracy of theperformance prediction mechanism for the entire population of TBIs. In our model, the decentralized blockchainnetwork is also used to address security concerns related to unauthorized access and data manipulation thereby ensuringtransparent and tamper-proof model updates. We evaluate the performance of our proposed PPFL-BC model by utilizingreal-world business incubation datasets. The simulation results show that our model outperforms the centralizedperformance prediction models in terms of accuracy, precision, recall, and F1-score. The results show that the proposedPPFL-BC model outperforms benchmark models with an accuracy of 84% and precision of 0.92, which shows theefficiency and reliability of our model in predicting and validating TBI success rates.
Advanced Metering Infrastructure (AMI) connects smart meters, data concentrator units, and utility controlcenters through persistent two-way communication. This architecture improves demand response and distributed-energymanagement, but it also exposes resource-constrained meters to replay, false-data injection, physical extraction, andlong-term key compromise. This article develops a formally verified and statistically evaluated lightweight AMIauthentication and key agreement protocol for resource-constrained smart-grid deployments. We first reconstruct theAMI authentication workflow as a four-message lightweight authenticated key exchange and map each entity, message,and key dependency to a smart-grid deployment model guided by NISTIR 7628 and IEC 62351. We then identifyreplay-within-window exposure, insufficient responder freshness, weak identity-to-key binding, missing keycompromise impersonation protection, and retrospective session-key recovery. To address these weaknesses, wepropose AMI-AKE, a transcript-bound protocol using ephemeral Curve25519 contributions, session identifiers, nonceand timestamp binding, binding signatures, and separate key-derivation function (KDF) outputs for encryption andintegrity. ProVerif-style verification queries and an extended Canetti-Krawczyk (eCK)-oriented game proof areprovided for mutual authentication, secrecy, forward secrecy, and key-compromise impersonation (KCI) resistance. AContiki-OS and ARM Cortex-M4 benchmark with 1,000 repeated trials reports 18.4 +/- 1.2 ms authentication latency,542 +/- 9.1 sessions/s throughput, and 99.2 +/- 0.4% false-data-injection detection under controlled prototype conditions.The proposed design replaces subjective security labels with objective metrics, confidence intervals, and a reproduciblesimulation plan for 1,000-10,000 smart meters.
The article discusses methods for averting irrational product placement in warehouses. This problem isrelevant for many manufacturing enterprises and trade organizations. The most common embodiments of the problemare unoccupied areas, or a lack of free storage space, difficulty in locating a specific product, and challenges in shippingit from the warehouse. All this leads to unnecessary costs for the business entity and hurts its financial and economicactivities. The author suggests an integrated approach to warehousing. It integrates both classical optimization modelsand iterative approval procedures for accounting for the human factor. The key criteria in this case are minimizing costs,the cargo flow in the warehouse, and maximizing the utilization factor of the usable area. The optimal placement ofgoods is to achieve maximum compression of their residues in the warehouse while minimizing their movement. Thepresence of two contradictory criteria makes the task a task of consistent optimization. The article discusses thepossibilities for solving the optimization problem when conflicting target criteria and differing preferences are present.We are using the example of a storage room with 16 racks for water heaters and similar equipment. As a result ofmatching optimization procedures, it was possible to reduce the average cost of moving goods by 7.1% and increase thefree warehouse area by 15 times. We performed the experiments over the seven days of the warehouse’s operation. Thepractical value of the research is that, through this approach, we find a compromise in conditions of conflicting opinionsand interests.
Remote sensing images are complex, which makes it difficult to interpret and generate semantically appropriate textual description. To get a semantically relevant description, it is important to identify complex objects and understand the contextual relationships between them. In such cases, deriving contextually accurate information while maintaining semantic coherence is challenging. Therefore, a specifically designed model architecture is required to generate semantically relevant descriptions. This paper discusses a deep learning-based approach to generate remote sensing image descriptions using an end-to-end encoder-decoder model with soft attention. The UC Merced (UCM) dataset is used for training, which includes multiple captions per image capturing various scene aspects. To further assess the robustness and generalizability of the proposed approach, its performance is additionally evaluated on more complex datasets such as RSCID and Sydney Captions. This study presents an end-to-end CNN–LSTM encoder–decoder framework enhanced with soft attention for semantic description generation from remote sensing imagery. The framework employs a VGG16 encoder to extract a 4096-dimensional visual feature vector, which is projected into a 256-dimensional representation and processed by a 256-unit LSTM decoder. The soft attention mechanism dynamically computes attention weights using the encoder features and decoder hidden state, enabling the model to emphasize relevant visual information during word generation. Multiple CNN encoders and learning rates are evaluated with LSTM decoders, both with and without attention, on the UCM, RSCID, and Sydney Caption datasets. At a learning rate of 0.0001, VGG16–LSTM with soft attention achieves BLEU-4 (B4) scores of 0.6636, 0.6636, and 0.5864 on the UCM, RSCID, and Sydney Caption datasets, respectively, compared with 0.1643, 0.1647, and 0.1745 for VGG16–LSTM without attention. The results demonstrate that soft attention substantially improves description generation by strengthening visual–linguistic alignment and enabling more contextually relevant and semantically coherent descriptions across datasets with varying scene complexity.
The paper deals with coalitions whose members are unselfish. Coalition members do their best to completethe arising tasks, and do not expect to receive a reward. A coalition member can be an entity such as a social orgovernmental organization, a military unit, or a complex technical device such as an autonomous robot, or any otherentity that has the capabilities, willingness, and ability to cooperate. The paper considers the non-redundant coalitions,which have only those coalition members without whom they cannot perform the tasks. In the paper, we consider onlythe situations when substitution of failed coalition members is impossible. A coalition tolerates the failure of itsmembers by using the surplus of coalition capabilities. In our research, coalition capabilities are understood as resourcesand services (e.g., materials, energy, power etc.). During the execution of tasks, one or more members of the coalitionmay fail. The paper uses the probability of the event that the coalition tolerates the failures of its members to evaluatethe coalition fault tolerance. A method is proposed for determining the coalition fault tolerance in the case of multiplemember failures. Complexity of the proposed method and its applicability are assessed.
This paper proposes IPAMS (Interview Performance Assessment using Gen AI), which is an AI-drivenplatform that automates interview evaluations using advanced technologies like Convolutional Neural Networks(CNN) to gain insights on facial emotions and expressions, Large Language Models (LLM) to generate and processinterview questions, YOLO (You Only Look Once) for real-time object detection, and APIs for speech-to-texttranscription and behavioral analysis. The system captures video responses and analyzes key elements such assentiment, speech patterns, body posture, and facial expressions, generating a detailed report. This report highlightsa candidate’s strengths and areas of improvement and is sent directly to their email with actionable insights. IPAMSmodernizes recruitment by providing unbiased assessments, saving time and resources for recruiters. For candidates,it offers a valuable mock interview tool, delivering feedback on technical skills, confidence, stress levels, andnonverbal communication. By combining cutting-edge AI and analytics, IPAMS delivers an efficient, objective, andinsightful solution for recruitment and self-assessment, benefiting all stakeholders in the interview process.
This research examines RAFA-BioAuth, a risk-adaptive, fairness-aware framework for mobile banking incases of presentations and facial occlusions. The proposed solution combines aspects of: Identity Similarity, PassivePresentation Attack Detection (PAD), Asymmetric Financial-Risk Estimation and Fairness Regularization at theIdentity Level. For evaluation purposes, all benchmark datasets were split into subject disjoint training, validation, andtesting sets. The threshold values from the validation set were used. Bootstrap resampling was employed to estimate thevariance. Monte Carlo simulations were performed to estimate the risk. The results showed that identity verification(AUC = 0.548) and PAD (AUC ≈ 0.55) were poor. Comparing RAFA to the AND rule resulted in FAR = 0.20, FRR =0.31, and expected risk = 340. On the other hand, the AND rule had lower FAR of 0.09, but increased FRR to 0.78.Finally, a conservative end-to-end approach produced an FRR of 0.686. Thus, our results indicate trade-offs rather thanproduction-readiness as we did not perform deployment, cross-device, or cross-dataset validations on our solutions. Wepresent the contributions of this research as being an interpretable integration and not a new algorithm.
The demand for data-driven insights in government has highlighted the importance of collective analytics. This study attempts to explore the key challenges of collective analytics in the context of Indian e-governance and the framework for addressing them. The study is based on a literature review, references to two cases, and expert views obtained from professionals involved with analytics solutions in government. In this study, analytics projects are considered as dashboard-based analytics. Based on the content analysis of expert responses, 14 key challenges of collective analytics in e-governance have been identified. The novelty of the present study is the focused exploration of challenges and their framework related to collective analytics in e-governance-a topic that received limited attention in the extant literature. This study brings forth the fact that unless the challenges of collective analytics in e-governance, including those related to data visualization, data quality, capacity building, technological capabilities, and inter-agency communications, are recognized, the implementation of collective analytics can be challenging. This study provides the basic understanding needed for data-driven governance through collective analytics. The output of the study will be helpful to the managers, e-governance experts, academicians, planners, and policymakers to understand the dynamics of collective analytics in government for handling discussed challenges well in advance. This study will also helpful to reduce the cost and time of the collective analytics project for effective decision-making.
Job Shop Scheduling Problem (JSSP) has become one of the key issues in a contemporary manufacturingsystem in which the task is to optimally schedule jobs to the machines to reduce the time and resources used inproduction. Good scheduling is critical in enhancing the productivity and competitiveness of manufacturing industries.In this research, Artificial Fish Swarm Optimization (AFSO) algorithm is used to optimize the JSSP in minimizingmakespan, total work load and maximum work load in machines. The AFSO strategy models the swarm behaviour offishes to search and forage the search space in an efficient manner to prevent its early convergence to local optima. Themodel incorporates a disturbed state in the world to improve the direction in search and the speed of convergence. Theeffectiveness of the suggested AFSO method is compared and tested with the traditional and sophisticated optimizationalgorithms like Particle Swarm Optimization (PSO) and Artificial Bee Colony (ABC) methods. The experimentalfindings prove that the offered technique provides better results in convergence rate and solution quality. The resultsprove that AFSO is a useful and promising method of solving complex problems in production system scheduling.
In the dairy industry, optimizing reproductive management is crucial for sustainable operations and enhancing animal welfare. The traditional manual detection methods usually miss many of the estrus incidences and hence have resulted in a 20-30% decline in conception rates and further massive economic losses.This paper presents an advanced framework integrating machine learning and Internet of Things (IoT) technologies to improve estruses detection in dairy cattle, thereby supporting efficient herd management and productivity. The proposed solution leverages a stacking model of Random Forest and Gradient Boosting Machine (GBM) algorithms to accurately identify estruses events, providing a reliable method for reproductive monitoring. The experimental evaluation yields accuracies of 92.1 % using RF, 92.3 % using GBM, and an improved 93.19 % when the stacking model is applied, along with improvements in precision of 94 and an F1-score of 94 %, reflecting its strength in complex behavioral pattern recognition. Rigorous evaluation across key performance metrics confirms the model’s high accuracy, underscoring its suitability for practical deployment. The system employs IoT-enabled smart collars equipped with temperature sensors, accelerometers, GPS, and RFID to gather real-time data on cattle health and reproductive status. By analyzing this data, the system delivers precise and timely insights into estruses cycles, enabling targeted breeding interventions and enhanced reproductive management. Data collected through the smart collars is securely stored in Google Firebase, facilitating efficient data archiving and rapid access via a user-friendly web application. The proposed integration of IoT, machine learning, and cloud computing presents a holistic, scalable, and economically viable solution for enhancing reproductive efficiency, animal welfare, and sustainable dairy management
Despite cloud computing's scalability and economy, energy efficiency, security, and equitable scheduling remain significant concerns. The traditional scheduling approach often fails to optimize execution time, energy consumption, and security concerns, resulting in less resource utilization and less secure systems. This paper proposes the Hybrid Bat-Genetic Algorithm (HBA-GA), which combines the Bat Algorithm for fast exploration with the Genetic Algorithm for accurate exploitation. This method reduces energy use while also reducing security risks like unauthorized access and data leaks. It uses Jain's Fairness Index (JFI) in order to ensure that workloads are evenly distributed and VM overload and conflicts are avoided. Based on simulations results, proposed HBA-GA improves energy efficiency while reducing security exposure and risk likelihood at the scheduling level by incorporating security-aware risk scoring into task–VM allocation decisions.
This research proposes the implementation of the subjective and objective weighting approach (SOWA) method as a new approach in determining the criteria weights that combines subjective assessments from experts and data-driven objective calculations. The criteria weights generated from the SOWA method are then used in various multi-criteria decision-making (MCDM) methods, such as simple additive weighting (SAW), technique for order preference by similarity to ideal solution (TOPSIS), multi-objective optimization on the basis of ratio analysis (MOORA), grey relational analysis (GRA), multi-attribute utility theory (MAUT), weighted aggregated sum product assessment (WASPAS), weighted product (WP), simple multi-attribute rating technique (SMART), multi-attributive border approximation area comparison (MABAC), and Multi-Attribute Ideal-Real Comparative Analysis (MAIRCA), to evaluate and rank alternatives. The research results show that the SOWA method is capable of producing balanced and representative weights, as well as consistent alternative rankings across MCDM methods. Sensitivity analysis of the ranking results indicates that all methods yield identical ranking results, signifying a high level of stability and reliability of the generated weights. These findings demonstrate that the SOWA method can serve as a solid foundation in decision support systems, particularly in the context of candidate selection or evaluation based on multiple criteria.
The rapid growth of mobile wallet usage has led to a sharp increase in fraudulent transactions, making fraud detection in portable wallets a pressing concern. Accurately detecting fraud is difficult because transaction data is complicated and unbalanced. Conventional rule-based systems are less flexible and frequently provide large false positive rates along with poor accuracy. Effective feature selection is crucial to the performance of Machine Learning (ML) models, notwithstanding their increased detection rates. Redundancy and noise are introduced by high-dimensional data, which lowers model performance and raises computing costs. The advantages of hybrid feature selection are frequently overlooked in current research, particularly when it comes to portable wallet fraud detection. By combining Random Forest Importance, LASSO Regression, Recursive Feature Elimination (RFE), and Mutual Information (MI) with resampling to solve class imbalance, this study fills that gap. Our approach provides a more reliable and effective solution for safe portable wallet fraud detection by removing superfluous features, increasing accuracy, and reducing computing cost. The model becomes faster and more effective when superfluous characteristics are eliminated because this reduces the computational effort. By concentrating just on the most instructive data, it increases accuracy. By addressing class imbalance and combining several selection strategies, the hybrid approach guarantees robustness. All things considered, this leads to a scalable and safe fraud detection system for transactions using mobile wallets. Our results show that a successful feature selection approach improves fraud detection accuracy, which in turn improves operational effectiveness and financial security.
The beginning of the fourth industrial revolution or Industry 4.0 has changed the concept of automation in industries by adopting the Internet of Things (IoT) in manufacturing, logistics, and production processes. The IoT is the digital foundation of Industry 4.0 that allows real-time monitoring, predictive maintenance, data-based decision making, and autonomous processes with in-between devices and smart sensors. This review explores the uses of the IoT in industrial automation through analyzing the enabling technologies, communication protocols, integration with cloud computing, wireless sensor networks, edge computing, artificial intelligence (AI) and machine learning (ML), as well as presenting important applications, current challenges, and future trends in smart industrial systems. Instead of considering the technologies separately, this paper takes the system-level viewpoint by integrating the way IoT architecture, communication protocol, intelligent analytics, and security controls collectively facilitate Industry 4.0 automation.
In the era of social media-driven communication, sarcasm poses a big challenge for the automated sentiment analysis systems, much more on platforms like Twitter, due to the brevity and often contextually ambiguous nature of the text. Misinterpretation of sarcastic content may degrade the reliability of downstream analytics, encompassing opinion mining and content moderation. To address this challenge, we propose, in this paper, a multi-modal transformer-based approach to sarcasm detection, which integrates textual and emoji information through the use of a cross-attention mechanism. The proposed model utilizes RoBERTa for the contextual processing of textual content to generate contextualized text embeddings, whereas emojis are encoded using Emoji-BERT to capture emoji-specific semantic and emotional cuing. A Gated-LSTM layer has been employed to model sequential dependencies among emojis, and a cross-attention mechanism dynamically aligns emoji representations with textual features for enhancing the sarcasm recognition capability. Later, these fused representations are passed to a fully connected classification layer for predicting sarcasm. For the evaluation of the performance of our proposed model against state-of-the-art results, standard metrics of evaluation have been considered. Experimental results demonstrate that the proposed approach outperforms several baseline and state-of-the-art models, with an accuracy of 92.5%, precision of 91.8%, recall of 93.2%, and an F1-score of 92.5%. From these results, we learn that jointly modeling textual and emoji modalities improves the performance of sarcasm detection in social media content. Also, these findings illustrate the potential of the suggested approach in improving sarcasm-aware sentiment analysis in the realm of social media analytics and automated content moderation systems.
This study proposes a hybrid quantum-classical framework for depression detection from social media text, integrating a frozen DistilBERT encoder with a variational quantum circuit (VQC)-based classification layer. The motivation stems from challenges in clinical NLP, including overfitting on limited datasets and high parameter overhead in conventional deep learning classifiers. Experiments are conducted on a balanced subset of the Reddit Self-Reported Depression Diagnosis (RSDD) dataset comprising 6,000 users. The proposed model is evaluated against classical baselines, including TF-IDF with logistic regression and a fine-tuned DistilBERT model. Results indicate that the hybrid approach achieves competitive performance, with an F1-score of 0.925 (±0.009) and improved recall (0.942 ± 0.015) compared to the classical DistilBERT baseline. Additionally, the quantum classification layer requires significantly fewer trainable parameters (72) compared to the classical dense head, demonstrating improved parameter efficiency at the classification stage. While the results suggest that variational quantum circuits can serve as an alternative non-linear classifier in low-data settings, the findings are based on simulation and require further validation on real quantum hardware. This work contributes to the emerging area of quantum natural language processing by providing an empirical evaluation of hybrid architectures on a real-world clinical text dataset.
The advent of (Internet of Things) IoT technologies has essentially transformed traditional houses into intelligent, equipped, and networked smart houses that serve to improve the quality in the lives of human beings with respect to security, energy efficiency, and comfort through massive automation, sensing, and remote control. However, with such a shift of paradigm, due to the diversity of devices, the limitation of resources, problems of interoperability, and a growing array of cyberthreats, opens up numerous avenues for security and privacy threats. This review attempts a holistic coverage of IoT-based smart home technologies and then provides a systematic classification of the security vulnerabilities from device, network, cloud, and application layers. The key threats include unauthorized access, data leakage, propagation of malware, denial of service, and exploits targeted against AI, with an analysis of their causes and occurrences in the real world. The paper undertakes a critical assessment of contemporary countermeasures, ranging from lightweight cryptographic protocols, AI-driven intrusion detection systems, blockchain-based authentication, privacy-preserving edge computing, and zero-trust frameworks. A comparative insight into each approach conversed with the views of the established literature draws out trade-offs between security efficacy, scalability, computational overheads, and user adoption. Based on a synthesis of the modern findings, continued gaps are identified, and future directions provided: including quantum-resistant encryption, interoperable standards, and user-centric security design, acting as the working platform or actionable directions for any researchers, developers, or policymakers in building of secure, resilient, and privacy preserving smart home ecosystem.
In recent years, the rapid advancement of machine learning (ML) has surpassed many expectations, and its application in the healthcare sector has emerged as one of the most fascinating areas of exploration. This thesis looks into whether machine learning can increase the precision and efficacy of breast cancer diagnosis. With the help of nine classification algorithms including Random Forest, XGBoost and MLP Classifier the given work intends to propose a reliable automatic solution for malignant and benign classification of breast tumor. The main idea of the project is the development of the Web based tool that would allow doctors and other medical practitioners to make quick decisions The MLP Classifier was found to be the optimal solution after its efficiency was evaluated based on the accuracy rate, and such parameters as precision rate, recall rate, and F1-score. This leads to development of a user friendly app; even those that would not originally consider themselves technical can easily operate the application. Apart from addressing the matter of high accuracy of diagnostics, the system shows the possibility of minimizing the rates of human factors and optimizing clinical decision. Seeking for that day when technology and human opinion will complement each other in the delivery of healthcare, our study neither only contributes to the growing literature on applying artificial intelligence in healthcare but also evolves the blueprint to integrate ML models in everyday practice.
In modern educational environments, particularly within computer laboratory settings in higher education institutions, the lack of effective real-time supervision and streamlined assessment processes presents a persistent challenge. Most current systems still rely on manual monitoring and evaluation, which are not only inefficient and time-intensive but also vulnerable to academic dishonesty, such as copy-paste behaviour during lab work. This study identifies and addresses this critical gap by proposing the development and implementation of an integrated real-time monitoring and assessment system tailored for use in academic computer labs. The proposed solution is a desktop-based application that incorporates four key features: Real-Time Viewer (RTV) for live monitoring of student activities, Block Inappropriate Websites (BIW) to restrict access to non-educational or harmful content, Manage Computer Time (MCT) to regulate system usage duration, and Form Learning Assessment (FLA) for digitalized and efficient performance evaluation. The development process followed the System Development Life Cycle (SDLC) framework, ensuring a structured approach across analysis, design, implementation, testing, and maintenance stages. Empirical testing involved a series of functional test cases simulating real-use conditions. All seven critical scenarios—such as input validation, session management, access control, and data deletion—were executed and passed successfully, indicating the system’s robustness and usability. In a pilot study conducted at Pekanbaru College of Technology, the application was tested among 30 students across multiple laboratory sessions. The results demonstrated a notable improvement in student engagement and learning performance. Quantitatively, students achieved learning assessment scores ranging from 84 to 96, with a calculated mean of 89.6 and a standard deviation of 4.1. These outcomes suggest that the introduction of automated, real-time monitoring significantly enhances not only instructional supervision but also the accuracy and fairness of learning assessments. This research contributes to the field by bridging the gap between digital classroom management and performance assessment in a higher education context. It introduces an innovative and practical approach for educators to maintain instructional quality while managing multiple learners in digital settings. Moreover, the findings provide empirical evidence supporting the integration of real-time supervision tools into educational systems to foster accountability, deter academic misconduct, and support data-driven instructional improvements.
Connected autonomous vehicles (CAVs) are reshaping mobility but remain vulnerable to technical, organizational, and regulatory risks. This study develops a hybrid multi criteria decision-making framework that integrates CRITID for objective weighting, Fuzzy BWM for expert uncertainty modeling, and VIKOR Swarm for adaptive compromise ranking. To enhance realism, four scenarios were constructed: scalability focused (A1), compliance & reliability focused (A2), resilient high performance ecosystem (A3), and organizational vulnerability focused (A4). Results show that Scenario A3 consistently outperforms others, achieving the lowest group utility shortfall, smallest individual regret, and most favorable compromise measure. Shapley Value sensitivity analysis confirmed cybersecurity and scalability as dominant criteria, while expert AI validation reinforced the robustness of A3’s ranking. Monte Carlo simulations further demonstrated stability underweight perturbations, with A3 retaining its top position in over 80% of runs. The study contributes a transparent, reproducible, and scenario based methodology for vehicular risk assessment, bridging technical and organizational dimensions. Limitations include reliance on static scenario design and expert elicitation, suggesting future work should incorporate dynamic data streams and edge AI for real time risk recalibration.
