Breast Cancer is considered one of the largest and deadliest diseases among women because of its complex nature. Moreover, the chances of survival are greater when detection by mammograms is carried out at an earlier stage. However, existing automated systems have been influenced by noisy conditions, variability in datasets, and inconsistent protocols for imaging modalities. This research presents a comprehensive, multi-phase strategy for the detection and classification of breast cancer using mammographic images. It integrates three major steps: preprocessing, segmentation, and classification. Advanced techniques of preprocessing include adaptive histogram equalization, Gaussian filtering and MobileNetV2-Based autoencoder denoising, achieving a notable improvement in image quality with an average PSNR of above 63 dB. Segmentation is carried out using an attention-guided modified U-Net architecture, yielding high segmentation accuracy with a Dice similarity coefficient of 98.75%. The classification phase is performed using a stacked ensemble learning approach, where the predictions of EfficientNetB0, DenseNet121, and ResNet50 are combined into a Logistic Regression (LR) meta-learner. The proposed framework is evaluated on combined public and clinical datasets (DDSM, MIAS, and clinical data) and achieved a classification accuracy of 99.91%, sensitivity of 100%, precision of 99.85%, F1-score of 99.92%, and an AUC of 1.0. Therefore, this research opened a new perspective that seems to be quite effective in integrating advanced deep learning techniques together with the ensemble method to enhance the robustness and accuracy in automated breast cancer diagnosis.

VM consolidation is an effective approach for reducing energy consumption in cloud data centres. The selection of VMs from under-loaded or overloaded machines and migrating them on effective hosts constitutes the process of VM consolidation. The algorithms in literature to select hosts for VM deployment are generally based on a single criterion. However, VM placement is a multi-criteria decision-making problem. In this paper, an attempt is made to design a host selection technique based on improved Preference Ranking Organisation Method for Enrichment Evaluations (PROMETHEE) for energy efficient VM deployment. The proposed selection policy uses multiple parameters to find the selection index of hosts. The selected hosts help to reduce both energy consumption and service level agreement violations. A case study-based approach is followed to validate the proposed VM deployment framework using real data, real hosts and VMs configuration. Results indicate the employability of the framework in real cloud environments.

Thyroid Nodule (TND) refers to an abnormal structure composed of cells within the thyroid gland. These nodules are mostly benign in nature, but a small portion of nodule might be cancerous. Therefore, early identification and classification plays a significant role. Ultrasonography (USG) is a key tool for the thyroid nodule identification. USG uses a high- frequency sound waves to obtain picture of the nodules. Deep Learning (DL) provides a cutting-edge result to the various Machine Learning (ML) techniques and computer vision tasks. In this article, Generative Adversarial Network-Alex based synthetic ultrasound images augmentation model is proposed for TND classification. The Generative Adversarial Network-Alex model works in four phases: (1) data acquisition, (2) pre-processing, (3) data augmentation using Generative Adversarial Network (GAN), (4) classification using AlexNet. The model is evaluated on public and collected datasets having 295 and 428 thyroid ultrasonography (USG) images. GAN technique for data augmentation and Grid Search Optimization (GSO) employed to identify the optimal set of hyperparameters (learning rate and optimizer) that maximize the performance of the AlexNet model. The proposed Generative Adversarial Network-Alex model has achieved an accuracy of 96.85

Workflow scheduling in heterogeneous multiprocessor systems becomes substantially harder when a global energy budget must be satisfied alongside performance goals. Existing energy-constrained schedulers share a common weakness: task-level energy allocation is computed independently of processor assignment, causing pre-assigned budgets to misalign with the energy-frequency characteristics of the processors eventually selected. Greedy runtime methods compound this by exhausting slack on early tasks, starving critical tasks later in the schedule. This paper presents Energy Restricted Workflow Scheduling (ERWS), a three-phase heuristic for dynamic-voltage-and-frequency-scaling-enabled heterogeneous systems that minimizes makespan under a hard global energy constraint. ERWS prioritizes tasks using upward ranking, distributes slack energy in proportion to each task’s minimum energy requirement, and selects processor-frequency pairs by earliest finish time subject to per-task energy limits. The method is evaluated on six real-world scientific workflows and compared against five state-of-the-art energy-constrained scheduling methods. Under the tightest budget tested, ERWS reduces normalized schedule length by an average of 39.3 % over the greedy baseline and 5–9 % over existing static pre-assignment methods; gains persist, though with narrower margins, as the budget relaxes. Friedman and Wilcoxon signed-rank tests confirm statistical significance across all workflows and all 30 pairwise comparisons.

Breast cancer remains a significant global health concern, necessitating effective early detection and prevention strategies. Despite the effectiveness of mammography in early breast cancer detection, challenges persist due to issues such as Low image quality, limitations of traditional segmentation methods, and suboptimal feature extraction techniques. This paper proposes a novel framework for detecting and classifying breast cancer from mammogram images based on a multilevel semantic segmentation approach that integrates GoogleNet and Multi Dilated Convolutions, hence named as GN-MDC. Apart from this, it is also expected that the improvement in segmentation accuracy with complicated anatomies like pectoral muscles will be enhanced according to the optimized feature selection attained by the WCO algorithm for classification performance. This integration enables the proposed model to accurately classify breast cancer as normal, benign, or malignant. The proposed framework provides better outcomes regarding accuracy, sensitivity, specificity, precision, false negative rate (FNR), f-measure and area under the curve (AUC). This work achieved (99.40 %, 99.53 %), (99.65 %, 99.80 %), (99.08 %, 99.22 %), (99.24 %, 99.32 %), (0.34 %, 0.24 %), (99.44 %, 99.56 %) and (99.37 %, 99.51 %) of accuracy, sensitivity, specificity, precision, FNR, f-measure and AUC on (collected, DDSM datasets) respectively. This study addresses significant challenges in breast cancer identification, presenting a robust solution leveraging advanced deep-learning techniques. The proposed Op-CNN framework contributes toward the overcoming of limits set by the existing methodologies by promising higher sensitivity and specificity with accuracy. Extensive experiments are conducted to prove the efficiency of the approach; it shows great potential in the domain of breast cancer detection and diagnosis.

Background: India is among the world’s largest producers of the potato. However, early blight and late blight caused by biotic stresses can result in considerable potato yield loss. Traditional methods of leaf disease detection are cumbersome, inefficient and take a long time. Timely and accurate classification of leaf disease can play a major role in increasing crop yield and agricultural sustainability. Methods: This study proposes an advanced PoLIVR approach for potato leaf illness classification using VGG-19 and Random Forest. This research used potato leaf images from a publicly available dataset. The raw images are pre-processed using grey scaling, denoising and bilateral filtering. VGG-19 is utilized to extract features from the pre-processed output of the pictures. Result: The proposed PoLIVR approach achieved 95.59% accuracy for illness classification with 96.41%, 91.44% and 93.64% precision, recall and F1-score. The obtained AUC values for the three classes were higher than 0.90, which suggests a strong positive power. The proposed approach also has a higher performance than the compared state-of-art work. These results highlight the potential of PoLIVR approach for timely and accurate leaf illness classification. Overall, the study will also serve as a reference tool for potato grower industry workers to identify diseases in different agricultural settings and facilitate crop protection decision-making.

Sarcastic posts are common on social media, where user’s express sentiments that differ from the literal meaning of their words. Accurately understanding the intended meaning behind these posts could greatly enhance the effectiveness of text analysis tools for social media, a topic that remains under-explored. This research introduces a bidirectional long and short term memory (BiLSTM) based encoder-decoder model incorporating an attention mechanism and SpaCy-based embeddings. The model processes sarcastic input text and generates their corresponding non-sarcastic interpretations. A key feature of the proposed model is its bidirectional architecture, enabling it to capture contextual information in both forward and backward directions. By utilizing an attention mechanism, the model focuses on relevant preceding and succeeding words to make accurate predictions. Additionally, external embeddings have been integrated into the model to enhance its performance, addressing limitations in existing sarcasm interpretation models. Which operate unidirectionally without leveraging external embeddings and attention mechanisms. The results demonstrate that the proposed model outperforms traditional unidirectional models in context generation, due to its attention-based bidirectional nature with external embedding for interpreting sarcasm.

Clustering is a well-known approach that is often used in a variety of domain such as data analytics, information retrieval, social mining, image analysis, etc. The literature has described numerous methods based on different clustering concepts; however, partitional clustering methods are commonly used due to their simplicity and ease of implementation. Conventional algorithms such as k-means suffer from some shortcomings like local optima, convergence rate, and population diversity. Heuristic approaches can address these shortfalls of clustering methods effectively. This work utilizes a recently developed meta-heuristic algorithm, called transient search optimization (TSO) to alleviate the issues of traditional clustering algorithms. The proposed work aims to solve the data clustering problem in more effectively and efficiently manner. A number of benchmark clustering datasets are used for assessing the proposed algorithm efficacy, and the outcomes are contrasted with standard clustering metrics like intra, rank and SD measures. The finding demonstrates that the TSO outperforms other algorithms based on intra and rank parameters.

Data clustering is a fundamental task in the field of machine learning which involves the partitioning of the datasets into meaningful groups. The traditional clustering algorithms often struggle with issues such as initial centroid sensitivity, slow convergence, and local optima trap. On the other side, meta-heuristic algorithms consist of innovative paradigms to handle these issues. Hence, this work introduces a new meta-heuristic algorithm, called transient search optimisation (TSO) to alleviate the issues of traditional clustering algorithms. Further, some enhancements are included in TSO algorithm to generate more optimal results. These improvements aim to make TSO more reliable for data clustering problems. The efficiency of the TSO is evaluated over benchmark datasets and results are compared using intra cluster distance, accuracy rate and detection rate parameters. The average accuracy rate and average detection rate of the proposed TSO algorithm are 6.94% and 6.48% higher respectively as compared to other algorithms.