Hybrid quantum classical algorithms offer a promising approach for exploiting Noisy Intermediate Scale Quantum devices; however, their practical deployment remains limited by excessive circuit depth, hardware noise, and inefficient utilization of quantum resources. This paper presents a Hybrid Variational Optimization (HVO) framework that reduces quantum algorithmic complexity by integrating adaptive circuit pruning with classical feedback-driven optimization and a unified Quantum Complexity Metric (QCM). Unlike prior adaptive variational approaches such as RE-ADAPT-VQA, the proposed framework explicitly optimizes the trade-off between gate count, fidelity, and qubit resources while remaining modular and hardware-agnostic. To ensure a fair and transparent evaluation, the proposed method is compared against a well-tuned shallow Variational Quantum Eigensolver (VQE), QAOA, and RE-ADAPT-VQA. Experimental validation on IBM Quantum simulators and real NISQ hardware demonstrates up to a 35% reduction in circuit complexity, a 30% reduction in execution time, and improved fidelity of 0.94. Furthermore, performance evaluation using accuracy, precision, recall, and F1-score confirms the robustness and reliability of the proposed framework. These results demonstrate that HVO provides a scalable and noise-resilient solution for near term quantum computing applications.

This work describes the design and the implementation of an AI document search chatbot using a Retrieval-Augmented Generation (RAG) architecture that combines generative modelling with information retrieval to provide an accurate and verifiable responses. The system supports PDF, DOCX, CSV, and TXT uploads, which are automatically processed, segmented, and indexed in a vector database for semantic search. It is implemented in Python using LangChain for orchestration, Facebook AI Similarity Search (FAISS) for vector retrieval, and multiple large-language-model Application Programming Interface (API), including OpenAI, Google Generative AI, and Hugging Face. For each query, embeddings are computed and the most relevant document segments are retrieved using cosine similarity before being passed to the selected model to ensure responses are grounded in user-supplied content. Owing to its modular design, the architecture is scalable, maintainable, and easily extensible for enterprise and academic knowledge-management applications.

A tremendous amount of data is constantly being generated by sources such as traffic and weather cameras, environmental sensors, digital infrastructure in city areas. In this situation, it has become a big challenge for the smart city managers to detect unusual events in real-time. In this paper, we have proposed a machine learning system based on the METR-LA city traffic dataset that detects abnormal conditions related to traffic and mobility. The proposed framework involves understanding the relationships between road segments by integrating temporal modeling with graph-based spatial learning and works accordingly to the dynamic changes that take place in the traffic flow. Using LSTM and transformer-based sequence modeling techniques, the system predicts normal traffic behavior and identifies deviations from that normal state - such as those associated with traffic congestion, accidents, or infrastructure faults. Experimental results show that, compared to traditional threshold-based and isolation-based methods, the proposed approach achieves greater accuracy, faster convergence, and lower false alarm rate. These results demonstrate that this approach can be effective in providing intelligent decision support to smart city traffic control systems.

Efficient anomaly detection and traffic monitoring are the advance features of Intelligent Transportation Systems (ITS) while existing techniques struggle in handling hazardous scenarios in traffic. Traditional detection methods based on bounding-box approach as discussed in earlier studies, show reduced effectiveness with variable vehicle types and complex traffic patterns deployed in heterogeneous traffic environment. This paper presents an advanced framework of computer vision for real-time traffic monitoring and anomaly detection in different traffic environments. The system utilizes YOLOv8 instance segmentation models that achieves pixel-level accuracy in vehicle detection while maintaining real-time processing capabilities. Using user-defined regions of interest (ROIs) in the paradigm of road-region focused detection eliminates background interference. The proposed framework implements pixel-level vehicle segmentation and supports simultaneous vehicle counting, traffic density analysis and substantially improves accuracy in anomaly detection by eliminating background noise. The proposed system was authenticated on urban traffic datasets across all analyzed frames (n=7) proving exceptional performance with consistent anomaly detection. The results displayed noteworthy enhancement in processing efficiency and detection accuracy in comparison to existing methodologies, including average occupancy percentages ranging from 63.31% to 71% and maximum occupancy reaching 93.74%.

Individual nutrition is a key issue in preventive care, but there is no reliable method of risk evaluation in the dieting process since the health data are heterogeneous, classes are disproportionate, and the probability estimates are inaccurate. It is suggested in the study that a hybrid deep learning architecture should be used to predict diet-quality and that includes transformer-based representation learning, gradient- boosted decision trees, latent-space oversampling, and calibrated ensemble modeling. The five actual health samples were combined into a single dataset of 8,737 valid cases of demographics, dietary intake, laboratory biomarkers, examination measurements and lifestyle questionnaires. A masked-feature pretraining plan was applied to acquire contextual nutritional representations, after which the embedding-level augmentation was taken to enhance the sensitivity of the minority-class. FT-Transformer, MLP and XGBoost models were combined using an optimized weighted ensemble with isotonic calibration to make them more reliable. The experimental results show the enhanced minority-class recall, the enhanced probability calibration, and the overall performance as opposed to individual models. These results suggest that the combination of calibrated learning methods with hybrid techniques can assist in efficient dietary risk assessment, as well as be used as a scalable base of individualized nutrition decision support systems.

Urban traffic congestion critically delays emergency service response times, leading to severe consequences. To address this, we propose a real-time intelligent system that enhances emergency vehicle transit through automated detection and dynamic route guidance. The system utilizes a YOLOv8 deep learning model to accurately identify ambulances, police cars, and fire trucks from live CCTV footage, demonstrating robust performance under challenging conditions such as varying lighting, visual noise, and dense traffic. Upon detection, the system dynamically computes the shortest viable route to the nearest relevant facility (e.g., a hospital) using the OpenRouteService (ORS) and Overpass API. To ensure rapid public cooperation, the suggested route is communicated through two primary channels: it is displayed on traffic signal LED screens and encoded into a QR code for direct mobile navigation. Our model achieves a high detection accuracy of mAP@0.5 = 0.984 and mAP@0.5:0.95 = 0.82. By integrating a state-of-the-art detection model with user-friendly routing interfaces, this framework significantly reduces emergency response delays, improves transit efficiency, and has the potential to save lives. Future work will focus on integrating live GPS data from emergency vehicles and expanding the system’s operational scope.

A complex neurodevelopmental learning disability, dyslexia is typified by ongoing difficulties with visual phonological mapping, orthographic processing, and reading. The fine-grained analysis of handwritten letter formation patterns can be a reliable indicator of dyslexic traits, according to recent developments in deep learning-based computer vision. However, most existing research focuses on coarse screening (dyslexic versus non-dyslexic) and lacks methods for continuously and interpretably measuring the severity of dyslexia. In order to quantify the severity of dyslexia through sequence modeling, this paper presents a hybrid deep learning pipeline that simultaneously uses a ResNet50-based embedding extractor and You Only Look Once (YOLOv5) for granular letter-level detection. Whole handwriting pages worth of extracted letter embeddings and spatial-structural patterns are used to train a custom sequence regression model. Dyslexia handwriting images with three representative error categories-Normal, Reversal, and Corrected letters-are used to test the suggested framework. According to the results, the suggested severity regression framework achieves an average RMSE of 0.0146 and Pearson correlations above 0.967 across folds, while YOLOv5 achieves a nearly perfect mAP50 (0.995) across all classes. These results indicate that, within this synthetic setting, severity scoring can be modeled as a continuous, geometry-aware regression task rather than a purely binary classification problem. The synthetic pages are made out of character samples from several sources including public handwriting data sets - making sure that there is a realistic variability at the letter level with programmatic control over layout and annotations. Severity scores in this study are computed as the sum of the reversal and corrected-letter rates per page. All experiments use programmatically generated pages derived from real character samples.

Chest X-ray classification studies often report endpoint performance without isolating how anatomical guidance, staged decisions, probability calibration, and error propagation influence the final prediction. CALM-CXR is a reproducible protocol that operationally couples lung-mask-conditioned regional expert modeling with separately trained normal-abnormal and bacterial-viral stages, stage-specific temperature scaling, calibration-partition threshold selection, and coherent three-class probability composition. Its methodological advance is the controlled, auditable coupling of these established components: each stage retains its own probability and decision parameters, final class probabilities remain mathematically coherent, and errors can be attributed to screening or subtype classification. The protocol was evaluated on 4,840 images using filename-derived group-disjoint train, validation, calibration, and test partitions. Across seeds 42, 52, and 62, the full protocol achieved accuracy 0.7920 (SD 0.0248), macro-F1 0.7999 (SD 0.0165), macro one-vs-rest AUROC 0.9063 (SD 0.0088), and macro PR-AUC 0.8616 (SD 0.0068). Removing lung-mask conditioning produced the largest mean decrease in accuracy (0.0234) and macro-F1 (0.0206) among the component ablations. A matched flat ConvNeXt-Tiny classifier achieved accuracy 0.7955 (SD 0.0135) and macro-F1 0.7982 (SD 0.0128), and paired comparisons did not establish consistent overall superiority of the hierarchical organization. CALM-CXR should therefore be interpreted as an auditable evaluation protocol, not as a new individual primitive or a clinically validated diagnostic system, independent external validation remains necessary.➢ Defines an auditable two-stage evaluation protocol that separates normal-abnormal screening from bacterial-viral subtyping and retains stage-specific probabilities and errors.➢ Tests lung-mask conditioning, learned regional routing, anatomy and routing-entropy penalties, temperature scaling, calibration-only thresholds, and hierarchical probability composition through controlled ablations.➢ Includes matched flat three-class baselines, paired statistical comparisons, calibration metrics, bootstrap uncertainty, and filename-derived group-overlap auditing.

Conventional weight loss programs often apply generalized nutritional guidelines, ignoring individual variations in metabolism and dietary response. Such limitations result in inconsistent progress and reduced long-term success rates. This research presents a data-driven framework to enhance personalized weight management by analyzing protein intake patterns using machine learning and deep learning models. The study employs data from the MyFitnessPal dataset and user-provided records, including diet logs, anthropometric parameters, and demographic details. Pre-processing is performed using a Target Encoding Wavelet Transform Filter (TEWTF) to remove noise and normalize heterogeneous data. Advanced algorithms such as Random Forest Generative Adversarial Networks (RFGAN) and Preference-Aware Inverse Optimization (PAIO) are implemented to create adaptive diet plans aligned with user preferences. Furthermore, Hierarchical Clustering combined with Multi-Layer Perceptron (HCMLP) is utilized to identify protein consumption trends and predict optimal intake levels for diverse user groups. The model dynamically refines nutritional recommendations based on individual behavior and physiological responses. Experimental evaluation shows consistent predictive accuracy, maintaining 95% performance across iterative testing from 100 to 900 posts. This framework demonstrates that machine learning can effectively translate nutritional data into personalized, real-time diet guidance, advancing the field of AI-driven health optimization.

Introduction: Having a noble cause, our work intends enhancing the quality of life for individuals who are visually impaired by effectively combining aural feedback systems with object detection technologies. By addressing practical problems, this incorporation improves the community’s general quality of life and safety. Objectives: Our primary objective is to provide individuals with visual impairments the necessary tools and information to navigate their surroundings effectively. By doing this, we hope to encourage inclusivity and enable them full involvement in daily activities. Methods: We used the You Only Look Once (YOLO) approach, namely YOLOv4, which is well known for its remarkable mean Average Precision (mAP) score of 92% and realtime object identification capabilities, to accomplish our goals. We had applied the 80 class names as well as bounding boxes in the COCO Dataset for effective item identification. This technology uses the relative distances among objects in order to offer visually impaired people more timely and reliable access to essential distance information. Results: We were able to create a system that can quickly and accurately detect items in real time when there are multiple objects by utilizing the COCO Dataset and the YOLOv4 approach.Furthermore, relative distance drive improves the user’s vision when identifying numerous objects. Conclusion: Our research indicates an important breakthrough in addressing challenges faced by people who are visually impaired. We developed a key strategy that encourages greater inclusivity and involvement in day-to-day activities while also improving mobility and safety through the application of innovative and state-of-the-art techniques.

Signature verification systems are critical in keeping the security of real individuals in a number of sensitive applications (e.g. banking or legal papers) Based on the nature of signatures of human beings it is very difficult to deal with the variability of human signatures and the problems in recognizing genuine signatures from fake ones in traditional methods. This work examines the usage of Multilayer Perceptron (MLP) Artificial Neural Networks (ANNs) for signature verification and investigates their capability in learning intricate patterns, allowing higher level classification. We propose a pioneering pipeline of MLP that works on static and dynamic signature data to provide feature extraction methods that accurately describe the signature features such as shape, velocity, pressure, timing in order to differentiate between genuine or fake signatures.MLP-ANN model is trained on the basis of an established signature sample set and subsequently verified using all the standard classification metrics like accuracy, precision, recall, and F1-score. The results show that MLP leads to much higher accuracy for signature verification than traditional machine learning efforts. Moreover, we elaborate about generalisation difficulty of neural model, dataset boundaries and preprocessing effects on performance. Using the results, in particular we point to the ability of MLP-ANNs to be applicable in simple ubiquitous verification systems on signatures but give directions for future work, e.g., integration with other deep learning models, as well adaptive to live/dead oriented systems.

In this study, we explore the efficacy of ResNet models in detecting handwritten text in the Telugu language, a domain with limited research. Our objective was to assess the adaptability and accuracy of these models in a challenging script recognition context. We employed a comprehensive dataset of Telugu handwritten characters, applying various ResNet architectures to evaluate their performance. The results reveal a significant improvement in detection accuracy, with the ResNet-50 model achieving the highest word accuracy rate of 84% surpassing previous benchmarks in Telugu text recognition. These findings not only demonstrate the robustness of ResNet models in handling complex scripts but also open avenues for further research in indigenous language processing. The study underscores the potential of deep learning models in enhancing text detection accuracy, marking a notable contribution to the field of optical character recognition for under-researched languages.

This research investigates the evolving advantages of diverse career trajectories while critically examining the current absence of standardized frameworks within the data science discipline. The pervasive dissemination of misinformation across the internet has precipitated an unprecedented surge in data generation, thereby necessitating the emergence of data mining as a crucial analytical paradigm. Over the forthcoming decade, the capability to extract contextually significant and substantively relevant insights from vast datasets is projected to become a fundamental determinant of progress across multiple domains.In particular, this study analyzes five case studies that collectively endeavor to establish a unified understanding of occupational structures within the data science ecosystem. The examination and utilization of data-driven insights to support empirical observation and informed decision-making constitute essential components, especially within the agricultural sector. Building upon these contextual analyses, the research delineates the distinct yet complementary roles of data engineers and data scientists in shaping effective data-centric strategies.To evaluate crop productivity, the study applies data mining methodologies with an emphasis on regression analysis. The estimation of individual crop yields is conducted through the identification of underlying patterns in parameter-specific variables and historical data records. This methodological approach facilitates the derivation of predictive insights, thereby contributing to enhanced agricultural output assessment and strategic planning.

A common shortcut in dialogue summarization is to truncate a conversation to its most recent turns, assuming this reduces cost without hurting summary quality. We test this assumption through a controlled context ablation study on DialogSum using DistilBART as the main model, with T5- small and BART-large for confirmation. Truncation consistently hurts quality: BERTScore drops by -0.0115 and ROUGE-L by -0.0705 (p < 0.0001), showing that removing early context acts as lossy compression that distorts meaning. Meanwhile, efficiency barely improves - shrinking input length by about 75% yields only a ~ 10% gain in throughput on an NVIDIA T4 GPU. These findings reveal that naive truncation is counterproductive; future dialogue summarization systems should adopt adaptive, content- aware context selection rather than blunt length reduction.

The enormous information overload caused by the exponential growth of digital content across businesses is making it more challenging to adopt effective automatic summarizing solutions. Abstractive text summarizing has grown to be an important area in NLP with the aim of creating concise, clear, and instructive summaries that go beyond basic source text extraction. Summarization systems continue to struggle with issues of semantic purity, factual correctness, and contextual alignment despite tremendous advancements made possible by deep learning and transformer-based architectures. The study uses the BART model, which has been refined on a sizable news dataset, to examine abstractive summarization with an emphasis on semantic and contextual alignment. The suggested system incorporates contextual alignment techniques and semantic-aware preprocessing to guarantee that generated summaries maintain coherence and fluency while preserving meaning fidelity. The main evaluation metric used in the study is ROUGE, which is supplemented by knowledge from more modern semantic similarity-based methodologies. When compared to baseline summarization systems, experimental results show significant gains in content selection, semantic consistency, and textual coherence, especially for longer inputs. The evaluation further highlights the benefits of incorporating semantic and contextual representations into training pipelines, showing that such improvements can reduce factual inconsistencies and improve the total informativeness of summaries that were produced.

Accurate perception in autonomous vehicles is critical for ensuring safety and reliability in complex driving environments. Conventional single-sensor or single-model approaches often fail in the presence of occlusions, low visibility, or dense traffic. This work introduces an advanced multi-sensor deep learning framework that integrates YOLOv8 for object detection, Cascade R-CNN for precise localization, DeepLabv3+ for semantic segmentation, BEVFusion for multi-modal data fusion, and DeepSORT for temporal tracking. By combining camera, LiDAR, and radar inputs, the framework establishes a robust pipeline capable of addressing adverse conditions. Experimental evaluations on benchmark datasets, including KITTI, nuScenes, and ApolloScape, demonstrate significant improvements in accuracy, recall, intersection-over-union, and multi-object tracking performance. The proposed architecture represents a reliable step toward safe and scalable autonomous driving systems.

Skin cancer, especially melanoma, remains a major global health concern due to its high mortality when not diagnosed early. Conventional diagnostic methods rely heavily on dermatological expertise, which may not be readily available in all regions. This paper presents an AI-powered system for early detection of skin cancer through dermatoscopic image analysis. The proposed method employs deep learning models-specifically CNNs, ResNet50, EfficientNetB0, and Vision Transformers (ViT)-to classify lesions as benign or malignant. By incorporating advanced preprocessing, transfer learning, and visual explanation techniques like Grad-CAM, the system enhances diagnostic accuracy and interpretability. Evaluated on publicly available datasets (HAM10000 and ISIC), the ViT-based model achieved 95.3% accuracy and a 95.0% F1 score, demonstrating high potential for real-time deployment in telemedicine and clinical settings.

In today's fast-paced business environment, middleware is essential for connecting applications, services, and databases, ensuring both security and reliability. This paper introduces DevSecOps, a pipeline framework designed to improve the security, automation, and rollback processes for managing cloud data. The primary goal of DevSecOps is to safeguard cloud data, maintain a seamless workflow, and ensure stable operations. At its core, DevSecOps combines continuous deployment, security, and system integration into a single, unified middleware pipeline. This creates a smoother, more efficient cloud data workflow. One of the standout features is the security verification module, which is built into every stage of the pipeline. It uses tools like OWASP, Sync, and SonarQube to detect vulnerabilities early, making security a priority from the start. The framework also automates key tasks like middleware configuration, orchestration, and containerization using technologies such as Kubernetes, Docker, and Ansible. This reduces the risk of human error, ensuring a more consistent and reliable environment. With DevSecOps, continuous monitoring and analysis are made possible through tools like Prometheus and Grafana. This enables real-time tracking of performance and early detection of any issues, helping to improve both security and overall efficiency. A key feature of DevSecOps is its rollback mechanism, which ensures stability during security processes by quickly recovering from failures. This minimizes downtime and prevents data loss, making it crucial for maintaining system reliability. Overall, DevSecOps fills the gaps left by traditional middleware systems by automating security, deployment, and workflow management. Experimental results show it reduces time complexity by 40% compared to existing models, with enhanced vulnerability detection.

This research presents a novel hybrid framework for heart disease prediction, integrating advanced data preprocessing with ensemble learning to enhance diagnostic accuracy. The methodology begins with rigorous data cleansing to ensure reliability, followed by Synthetic Minority Over-sampling Technique (SMOTE) to balance class distribution. Outliers are identified and mitigated using the Z-score method, preserving data integrity. A unique Recursive Hybrid Feature Extraction (RHFE) strategy, combining filter and wrapper techniques, optimizes feature selection by reducing multicollinearity and enhancing model efficiency. Key predictive markers include age, chest pain type, maximum heart rate, ST depression induced by exercise, and major vessel count via fluoroscopy. The refined dataset is used to train an CatBoost -based ensemble model, achieving remarkable performance with 94.2% accuracy, 93.5% precision, 94% recall, and an outstanding ROC AUC score of 0.98. These results highlight the model's robustness and its potential for real-world clinical implementation in early heart disease detection and risk assessment.

Non-Alcoholic Fatty Liver Disease (NAFLD) has emerged as one of the most prevalent liver disorders globally, affecting nearly one-third of the population, with particularly high incidence rates in countries like the UK. Despite its widespread occurrence, accurate estimation of its prevalence remains a challenge. Earlystage NAFLD, typically characterized by simple steatosis, can silently progress to more severe conditions such as non-alcoholic steatohepatitis (NASH), fibrosis, and cirrhosis if left untreated. This progression significantly compromises liver function and increases the risk of cardiovascular complications. However, current diagnostic methods, including magnetic resonance spectroscopy and ultrasound imaging, are often limited by cost, accessibility, and diagnostic specificity. Given the clinical urgency and the limitations of conventional diagnostics, this study addresses the critical need for an accessible and accurate method to detect early-stage liver disease—specifically, to predict NASH within the NAFLD spectrum. We propose a machine learning-based approach that leverages clinical and pathological data, including blood parameters and ultrasound-derived tissue characteristics, to support early detection. Using a dataset of 181 patients, we applied preprocessing techniques such as normalization and categorical encoding to prepare the data for modelling. Features such as integrated backscatter (IB), Q-factor, and homogeneity factor (HF) were extracted to quantify liver tissue characteristics. Support Vector Machine (SVM), chosen for its balance of simplicity and efficiency in handling high-dimensional datasets, was employed for classification and regression tasks. Experimental validation using Python-based implementations demonstrated the model's effectiveness, achieving an average accuracy of 89.95% across both clinical and imaging-derived datasets. This study underscores the potential of machine learning in improving early diagnosis of liver diseases and reducing their long-term clinical burden.

Outliers are considered as unexpected things observed while analyzing the data. Investigators found that the prediction and identification of outliers are extremely complex. Generally, a stream is measured as an unbounded data source executed promptly, and this research provides a novel way of predicting the outliers over the incoming data. Here, the incoming data is acquired from the hospital to validate the patients' records. There are higher chances of outliers over the incoming data based on the density of the arrival data. The execution of outlier prediction is performed independently with the integration of LSTM (Long Short Term Memory) over the stacked CNN model. The outlier detection process constantly measures the incoming data from the emotional input as an outlier or inlier. Here, the data reconstruction is achieved with the auto-Regressive model, and the prediction model considers the outliers to construct the training data. Various hidden representation acquired from the stacked model is considered for outlier prediction, and the experimental results demonstrate that the anticipated model shows superior prediction accuracy, specificity, sensitivity and F1 score. The proposed model is best-suited for prediction as the deep learning approaches perform well over complex applications and acquire superior results. The simulation is done in MATLAB 2016b, and the performance metrics show a better trade-off than the other approaches.

Dyslexia, a neurodevelopment condition impacting reading and language abilities, presents notable difficulties in promptly identifying and implementing effective interventions Traditional methods for diagnosing dyslexia often rely on subjective assessments and standardized tests, leading to delays in recognition and support. This paper offers an extensive examination of how deep learning techniques are applied in the domain of detecting and intervening in dyslexia. The integration of deep learning algorithms into dyslexia research offers promising avenues for more accurate and timely identification of individuals at risk. By leveraging neural networks and advanced machine learning models, researchers have begun to explore novel approaches that analyze linguistic patterns, eye-tracking data, brain imaging, and behavioral markers associated with dyslexia. Furthermore, this paper discusses the potential of deep learning in tailoring personalized interventions for individuals with dyslexia. These interventions aim to adapt to the specific learning needs of each individual, providing targeted support and enhancing the effectiveness of remediation strategies. While highlighting the advancements made in utilizing deep learning for dyslexia, this review also addresses challenges, including data scarcity, model interpretability, and ethical considerations. Additionally, it proposes future research directions that emphasize collaborative efforts among researchers, educators, and technology developers to foster the development of robust and accessible tools for dyslexia assessment and intervention.

This article describes the architecture and system design for assisting blind people in navigating freely inside an enclosed environment, such as the home or the outdoors. Thus, the proposed technology uses IoT technology and emerging techniques for machine learning to provide high-tech cane functionality that allows visually impaired navigators to walk independently. It also includes mobile applications to safeguard visually impaired persons and allow guardians to observe them. The proposed in this study system is intended to identify and classify any obstacles within a defined distance using machine learning. In this connection, an indoor and outdoor architecture on YOLO v3 is implemented for its detection technique, and multi-layer perceptron (MLP) neural network technology supports this framework. Based on the detection and classification, YOLO v3 and MLP are crucial for their accuracy.

Optical character recognition (OCR) technology is indispensable for converting and analyzing text from various sources into a format that is editable and searchable. Telugu handwriting presents notable challenges due to the resemblance of characters, the extensive character set, and the need to segment overlapping characters. To segment the overlapping characters, we assess the width of small characters within a word and segment the overlapping characters accordingly. This method is well suited for the segmentation of overlapping compound characters. To address the recognition of similar characters with less training periods we have used ResNet 18 and SqueezeNet models which have achieved character recognition rates of 95% and 94% respectively.

Recently, deep learning has been used in enormous successful applications, specifically considering medical applications. Especially, a huge number of data is captured through the Internet of Things (IoT) based devices related to healthcare systems. Moreover, the given captured data are real-time and unstructured. However, the existing approaches failed to reach a better accuracy rate, and the processing time needed to be lower. This work considers the medical database for accessing the patient's record to determine the outliers over the dataset. Based on this successful analysis, a novel approach is proposed where some feasible and robust features are extracted to acquire the emotional variations for various ways of expression. Here, a novel dense-Convolutional Neural Network (CNN) with ResNet (CNN-RN) extracts features from patients', while for establishing visual modality, deep residual network layers are used. The significance of feature extraction is less sensitive during outlier prediction while modeling the context. To handle these issues, this dense network model is used for training the network in an end-to-end manner by correlating the significance of CNN and RN of every stream and outperforming the overall approach. Here, MATLAB 2020b is used for simulation purposes, and the model outperforms various prevailing methods for consistent prediction. Some performance metrics include detection accuracy, F1-score, recall, MCC, p-value, etc. Based on this evaluation, the experimental results attained are superior to other approaches.

The explanation of the identified outliers is frequently provided to the users when other techniques are used to detect the outliers that are suggested in the study. This situation is quite challenging for the users to take the relevant actions that are regarded with the identified outliers as the outcome. If the outliers are detected, then the explanations need to be provided with outliers to reduce the challenge. Some surveys are available for the outlier interpretation, and some surveys are available to detect an outlier. The survey presents an outlier interpretation in the proposed system where useful knowledge is obtained from anomalous data to fill the gap and describe the data. The types of outlier explanations are defined, and the difficulties are discussed while creating every type. The existing outlier interpretation approaches are reviewed, and the processes of mentioning the difficulties are discussed. The applications of outlier interpretation are discussed, and the existing techniques are utilised for the determination of outlier explanations are reviewed. Further, the possible future research ways are discussed.

Proliferation of genomic, diagnostic, medical, and other forms of biological data resulted in categorizing of biological data as bigdata. The low-cost sequencing machinery, even in small research labs, is generating large volumes of data which now needs to be mined for useful biological features and knowledge. In this paper, we have used a NoSQL approach to handle the repeat information of the entire human genome. A total of 12 million repeats have been extracted from the entire human genome and have been stored using MongoDB, a popular NoSQL database. A web application has been developed to query data from the database at ease. It is evident that bioinformaticians tend to shift their database development approach from traditional relational model to novel approaches like NoSQL in order to handle the massive amounts of biological data.

Sixteen novel orsellinic esters (6a-l, 7a-d) along with four lecanoric acid related depsides (3a-c, 4) were synthesized and confirmed their structures by spectroscopic data (1H, 13C & HRMS). The synthesized compounds were evaluated for their in vitro α-glucosidase (Saccharomyces cerevisiae) inhibitory potential. Among the tested compounds, 3c (IC50: 140.9 μM) and 6c (IC50: 203.9 μM) displayed potent α-glucosidase inhibitory activity and found more active than the standard drug acarbose (IC50: 686.6 μM). Both the test compounds were subjected to in vivo antihyperglycemic activity using sucrose loaded model in Wistar rats and found compound 3c exhibited significant reduction in glucose levels.

A total of twenty-two novel coumarin triazole hybrids (4a-4k and 6a-6k) were synthesized from orcinol in good to excellent yields of 70–94%. The structures of all the synthesized compounds were elucidated by spectroscopic techniques such as 1H NMR, 13C NMR, and HRMS. The anti-inflammatory potential of synthesized compounds was investigated against the proinflammatory cytokine, TNF-α on U937 cell line and compounds 4d, 4j, and 6j were found to exhibit promising anti-inflammatory activity. These three compounds were further screened against TNF-α on LPS-stimulated RAW 264.7 cells, which confirm their anti-inflammatory potential. Furthermore, the above said active compounds were tested for their inhibitory effect on RANKL-induced osteoclastogenesis in RAW 264.7 cells by using tartrate resistant acid phosphatase (TRAP) staining assay at 10 µM. Molecular mechanism studies demonstrated that compound 4d exhibited dose dependent inhibition of RANKL-induced osteoclastogenesis by suppression of the NF-kB pathway. Thus, compound 4d is a promising candidate for further optimization to develop as a potent anti-osteoporotic agent.