Volume XI, Issue II

Title– A NOVEL DEEP LEARNING MODELS FOR EMERGENCY VEHICLE DETECTION

Author– HAIQING KI LIU

Abstract

The essence of sound events, manifested in their temporal and spectral structure within the time-frequency domain, forms the core of an evolving field focused on analyzing and categorizing acoustic environments through recorded sound. By employing convolutional layers, this study efficiently extracts high-level features that remain invariant to shifts in this domain. Our investigation centers on the detection of emergency vehicles. We explore three distinct deep neural network (DNN) architectures – dense layer, Convolutional Neural Network (CNN), and Recurrent Neural Network (RNN) – with varying configurations and parameters. Subsequently, we devise an ensemble model by meticulously selecting optimal models through experimental testing across various configurations, coupled with hyper-parameter tuning. This ensemble model achieves the pinnacle accuracy of 98.7%, surpassing the standalone RNN model, which achieves 94.5% accuracy. Additionally, we conduct a thorough performance analysis, comparing the deep learning models with a spectrum of machine learning algorithms, including Perceptron, Support Vector Machine (SVM), and decision trees.

Index Term– Audio recognition; CNN DNN; emergency vehicle detection; MFCC; RNN; siren sound.

DOI- 10.30696/IJEEA.XI.II.2023. 01-11

Reference to this paper should be made as follows:  HAIQING KI LIU, (2023), “A Novel Deep Learning Models For Emergency Vehicle Detection” Int. J. Electronics Engineering and Applications, Vol. 11, Issue II, pp. 01-11.

Title– AN INVOTATIVE DEEP LEARNING MODEL FOR GENDER-BASED CLASSIFICATION OF CANCER PATIENT

Author– XIAOYU LI UYYANG

Abstract

The advent of the Third computing platform, integrating Social, Mobility, Analytics, and Cloud (SMAC), has ushered in an era of unprecedented data generation across diverse domains like healthcare, finance, transportation, and cybersecurity. This surge in data, termed Big Data, poses challenges due to its unstructured and imbalanced nature, prompting the need for advanced analytical approaches. Deep Learning, rooted in artificial neural networks, has emerged as a powerful tool for handling the complexities of Big Data. Its ability to learn hierarchical representations of features enables it to extract intricate patterns, making it well-suited for various real-world challenges. In healthcare, for instance, Deep Learning algorithms like Neural Networks with Dropout and Random Forest have shown promise in classifying Medicare beneficiaries based on different scenarios. In one scenario, focusing on cancer-affected beneficiaries, the Deep Learning Neural Network with Dropout achieved impressive sensitivity, specificity, and accuracy scores of 99.17%, 97.68%, and 98.8%, respectively. This underscores its ability to discern complex patterns crucial for patient care. Techniques like Grid Search facilitate the identification of the most effective classifier configuration, enhancing predictive accuracy and robustness. Overall, the application of Deep Learning alongside traditional techniques offers significant promise for extracting valuable insights from Big Data across various domains.

Index Term– Deep Learning; DLNNWD; SMAC; Random Forest, SDG.

DOI- 10.30696/IJEEA.XI.II.2023. 12-22

Reference to this paper should be made as follows:  XIAOYU LI UYYANG, (2023), “An Invotative Deep Learning Model For Gender-Based Classification Of Cancer Patient” Int. J. Electronics Engineering and Applications, Vol. 11, Issue II, pp. 12-22.