BreastHybridNet: A Hybrid Deep Learning Framework for Breast Cancer Diagnosis Using Mammogram Images

International Journal of Computational and Experimental Science and Engineering, Год журнала: 2025, Номер 11(1)

Опубликована: Янв. 25, 2025

As a common malignancy in females, breast cancer represents one of the most serious threats to female's life, which is also closely associated with Sustainable Development Goal 3 (SDG 3) United Nations for keeping healthy lives and promoting well-being all people. Breast accounts highest number mortality early diagnosis key reducing disease-specific general. Current methods struggle accurately localize important regions, model sequential dependencies, or combine different features despite considerable improvements artificial intelligence deep learning domains. They prevent diagnostic frameworks from being reliable scalable, especially low-resourced healthcare settings. This study proposes novel hybrid framework, BreastHybridNet, using mammogram images tackle these mutual challenges. The proposed framework combines pre-trained CNN backbone feature extraction, spatial attention mechanism automatically highlight image area, contains signature patterns carrying information, BiLSTM layer obtain dependencies features, fusion strategy process complementarily. Experimental results show that accuracy 98.30%, outperforms state-of-the-art LMHistNet, BreastMultiNet, DOTNet 2.0 extent quantitatively. BreastHybridNet works towards feasibility interpretability scalability on existing systems while contributing worldwide efforts alleviate cancer-related cost-efficient lenses. highlights need AI-enabled solutions contribute accessing technologies screening.

Язык: Английский

---

Innovative Computational Intelligence Frameworks for Complex Problem Solving and Optimization

International Journal of Computational and Experimental Science and Engineering, Год журнала: 2025, Номер 11(1)

Опубликована: Янв. 9, 2025

The rapid advancement of computational intelligence (CI) techniques has enabled the development highly efficient frameworks for solving complex optimization problems across various domains, including engineering, healthcare, and industrial systems. This paper presents innovative that integrate advanced algorithms such as Quantum-Inspired Evolutionary Algorithms (QIEA), Hybrid Metaheuristics, Deep Learning-based models. These aim to address challenges by improving convergence rates, solution accuracy, efficiency. In context a framework was successfully used predict optimal treatment plans cancer patients, achieving 92% accuracy rate in classification tasks. proposed demonstrate potential addressing broad spectrum problems, from resource allocation smart grids dynamic scheduling manufacturing integration cutting-edge CI methods offers promising future optimizing performance real-world wide range industries.

Язык: Английский

---

Metaheuristic-Driven Optimization for Efficient Resource Allocation in Cloud Environments

International Journal of Computational and Experimental Science and Engineering, Год журнала: 2025, Номер 11(1)

Опубликована: Янв. 7, 2025

Intrusion Detection Systems (IDS) play a pivotal role in safeguarding networks against evolving cyber threats. This research focuses on enhancing the performance of IDS using deep learning models, specifically XAI, LSTM, CNN, and GRU, evaluated NSL-KDD dataset. The dataset addresses limitations earlier benchmarks by eliminating redundancies balancing classes. A robust preprocessing pipeline, including normalization, one-hot encoding, feature selection, was employed to optimize model inputs. Performance metrics such as Precision, Recall, F1-Score, Accuracy were used evaluate models across five attack categories: DoS, Probe, R2L, U2R, Normal. Results indicate that XAI consistently outperformed other achieving highest accuracy (91.2%) Precision (91.5%) post-BAT optimization. Comparative analyses confusion matrices protocol distributions revealed dominance DoS attacks highlighted specific challenges with R2L U2R study demonstrates effectiveness optimized detecting complex attacks, paving way for adaptive solutions.

Язык: Английский

---

A Context-Aware Content Recommendation Engine for Personalized Learning using Hybrid Reinforcement Learning Technique

International Journal of Computational and Experimental Science and Engineering, Год журнала: 2025, Номер 11(1)

Опубликована: Фев. 5, 2025

In the evolving landscape of e-learning, delivering personalized content that aligns with learners' needs and preferences is crucial. This study proposes a Context-Aware Content Recommendation Engine (CACRE) utilizes Hybrid Reinforcement Learning (HRL) technique to optimize learning experiences. The engine incorporates contextual data, such as pace, preferences, performance, deliver tailored recommendations. proposed HRL model combines Deep Q-Learning for dynamic selection Policy Gradient Methods adapt individual trajectories. Experimental results demonstrate significant improvements in learner engagement, relevance, knowledge retention. approach underscores potential context-aware recommendation systems revolutionize education by fostering adaptive interactive environments.

Язык: Английский

---

Dynamic Task Weighting Mechanism for a Task-Aware Approach to Mitigating Catastrophic Forgetting

International Journal of Computational and Experimental Science and Engineering, Год журнала: 2025, Номер 11(1)

Опубликована: Фев. 4, 2025

Catastrophic forgetting is still a big issue in sequential learning and particular for Natural Language Processing (NLP) models that tend to forget knowledge encoded previous tasks when new targets. To do this, we present Dynamic Task Weighting Mechanism which forms part of the Adaptive Knowledge Consolidation (AKC) framework. Our method dynamically adjust retention task similarity specific performance, while contrasted static regularization approaches such as Elastic Weight (EWC) Synaptic Intelligence (SI). This mechanism proposed involves computing embeddings with pre-trained BERT quantifying their from cosine similarity. complete above, compute score merged normalized performance metrics accuracy, F1 form an importance score. The model trades adaptability order retain previously learned by prioritizing important minimizing interference other unrelated tasks. We show our substantially mitigates results accuracy improvements on extensive experiments standard NLP benchmarks GLUE, AG News, SQuAD. Among baseline methods (EWC, SI, GEM), also has highest average 86.7% least amount 6.2%.

Язык: Английский

---

Application of Convolutional Neural Networks and Rolling Guidance Filter in Image Fusion for Detecting Brain Tumors

International Journal of Computational and Experimental Science and Engineering, Год журнала: 2025, Номер 11(1)

Опубликована: Фев. 4, 2025

Medical image fusion is the technique of integrating images from several medical imaging modalities without causing any distortion or information loss. By preserving every feature in fused image, it increases value for diagnosis and treatment conditions. A novel mechanism multimodal data sets proposed this paper. Each source smoothened using cross guided filter initial step. Guided output further to remove fine structures rolling guidance filter. Then details (high frequency) each are extracted by subtracting corresponding image. These fed convolutional neural networks obtain decision maps. Finally based on map maximum rule combination. We assessed performance our suggested methodology pairs datasets that accessible general public. According quantitative evaluation, recommended strategy improves average IE 12.4%, MI 41.8%, SF 21.4%, SD 22.81%, MSSIM 31.1%, 39% when compared existing methods, which makes appropriate use field accurate diagnosis.

Язык: Английский

---

Automated Diagnosis of Cancer Disease with Human Tissues using Haralick Texture Features and Deep Learning Techniques

International Journal of Computational and Experimental Science and Engineering, Год журнала: 2025, Номер 11(1)

Опубликована: Март 7, 2025

The increasing use of automated cancer diagnosis based on histopathological images is significant because it likely to increase the accuracy and decrease workload pathologists. This research introduces a hybrid methodology that integrates Haralick texture features with deep learning strategies improve identification in human tissue specimens. features, obtained from Gray-Level Co-Occurrence Matrix (GLCM), offer essential information regarding spatial relationships textural characteristics present samples, which frequently signal presence cancerous alterations. integration these interpretable convolutional neural networks (CNNs) makes our approach strengths both traditional analysis learning's ability learn complex patterns. will process raw image data leading powerful model that, hopefully, better classification along interpretability. These handcrafted capturing like contrast, correlation, energy, homogeneity, provide differences classify between normal cells abnormal ones. Experimental results were presented distinguishing non-cancerous tissues high accuracy. diagnostic efficiency was also enhanced while at same time providing reliable scalable tool may assist pathologists during clinical decision-making, consequently leads efficient patient care.

Язык: Английский

---

Comparative Evaluation of Feature Selection Techniques and Machine Learning Algorithms for Alzheimer's Disease Staging

International Journal of Computational and Experimental Science and Engineering, Год журнала: 2025, Номер 11(2)

Опубликована: Март 21, 2025

Dementia encompasses a range of brain disorders characterized by cognitive decline, with memory loss as hallmark symptom. Alzheimer's disease (AD), the most common form dementia, progressively affects functions, leading to severe loss. Early and accurate detection AD is essential for timely intervention, preventing further neuronal damage, improving patient outcomes. This study employs machine learning (ML) techniques, feature selection methods, texture analysis enhance diagnosis. By systematically evaluating various techniques Principal Component Analysis (PCA) in conjunction multiple ML algorithms, identifies effective approach classifying stages. The integration texture-based features models demonstrates significant improvement distinguishing Cognitive Normal, Mild Impairment, These findings highlight clinical significance combining early diagnosis, facilitating more precise classification contributing personalized treatment strategies.

Язык: Английский

---

Optimizing Energy-Efficient Task Offloading in Edge Computing: A Hybrid AI-Based Approach

International Journal of Computational and Experimental Science and Engineering, Год журнала: 2025, Номер 11(2)

Опубликована: Март 23, 2025

Edge computing has emerged as a pivotal technology for managing computational workloads in latency-sensitive applications by offloading tasks from resource-constrained Internet of Things (IoT) devices to nearby edge servers. However, optimizing task while ensuring energy efficiency remains significant challenge. This paper proposes Hybrid AI-Based Task Offloading (HATO) model, integrating Reinforcement Learning (RL) with Deep Neural Networks (DNNs) dynamically allocate resources minimizing consumption. The HATO framework formulates multi-objective optimization problem, considering factors such device workload, network latency, server availability, and constraints. Experimental evaluations demonstrate that the proposed model achieves 27.3% reduction consumption, 19.6% improvement completion time, 31.2% enhancement overall utilization compared conventional heuristic-based methods. reinforcement learning module adapts strategies real-time, optimal load balancing latency. Approach outperforms baseline models diverse scenarios, making it scalable efficient solution next-generation IoT applications.

Язык: Английский

---