Convolutional Neural Networks (CNNs) are a class of deep neural networks that have proven highly effective in areas such as image and video recognition. CNNs typically include several types of layers, such as convolutional layers, activation layers, pooling layers, and fully connected layers, all of which contribute to the network’s ability to recognize patterns and features. The pooling layer, which often follows the convolutional layer, is crucial for reducing computational complexity by performing down-sampling while maintaining essential features. This layer’s role in balancing the symmetry of information across the network is vital for optimal performance. However, the choice of pooling method is often based on intuition, which can lead to less accurate or efficient results. This research compares various standard pooling methods (MAX and AVERAGE pooling) on standard datasets (MNIST, CIFAR-10, and CIFAR-100) to determine the most effective approach in preserving detail, performance, and overall computational efficiency while maintaining the symmetry necessary for robust CNN performance.

Colon cancer remains a leading cause of cancer-related mortality globally, necessitating early and accurate diagnosis to improve patient outcomes. Traditional diagnostic methods rely heavily on manual interpretation by pathologists, which can result in inaccuracies and delays in treatment. This study proposes an innovative, automated approach to colon cancer diagnosis by integrating advanced machine learning techniques with deep learning architectures. We employed EfficientNet, a state-of-the-art convolutional neural network, to extract intricate features from histopathological images, alongside the Non-dominated Sorting Genetic Algorithm II for optimal feature selection. This hybrid approach significantly enhances diagnostic performance while reducing computational complexity. The model was evaluated using five diverse datasets: Colon Cancer Histopathological Images, Kvasir, Kvasir-SEG, Hyper-Kvasir, and Endotect. The results indicate that our method outperforms traditional models such as CNN, AlexNet, ResNet, and GoogleNet, achieving an accuracy of 99.97% on the Colon Cancer Histopathological Images dataset. These findings suggest that this novel approach can substantially enhance early detection and diagnosis of colon cancer, providing a scalable solution to current diagnostic challenges. Ultimately, our study lays the groundwork for future advancements in automated cancer diagnostics, contributing to improved patient outcomes and more efficient healthcare delivery. The code and dataset for reproducing these results are publicly accessible at https://github.com/Noushin-Saba/ColonCancerDetectionandDiagnosis.

The Internet of Things (IoT) is a concept that involves integrating diverse objects to enable seamless interaction between real-world and virtual entities. IoT is now connecting the physical world to networks. IoT devices can sense, process, transmit, and store data collected from the physical world. However, these devices are resource-constrained, creating significant security vulnerabilities in many IoT applications. Implementing effective security measures on such devices is challenging without compromising their performance or potentially causing damage. Consequently, there is a substantial gap between the security capabilities of current IoT devices and their security requirements. Computer security principles, namely Confidentiality, Integrity, and Availability (CIA), can be compromised by malicious intrusions or attacks on computers and information databases. This study proposed and compared 1D CNN and XGBoost for detecting malicious attacks in the IoT environment. The proposed techniques were evaluated on the five variants of NetFlow datasets. The experiments shows that the proposed techniques outperform the Ensemble Tree classifier, achieving better performance in binary and multi-class classification. The results for 1D CNN and XGBoost were compared on the basis of F1 measure, AUC, recall, correctness value, and ppv.. The comparison shows that XGBoost is the better-performing model across the NetFlow datasets. XGBoost's ability to capture complex patterns and optimize the classification task makes it robust and effective.