Performance Comparison of Multiple ANN Optimizer on IoT-enabled Sensor Fire Dataset

In today's world, fires in homes and commercial places are a serious problem that can harm the local environment as well as jeopardize people's property and lives. This study predicts the sensor dataset gained from an integrated sensor framework with an artificial neural network. The major goal of this research was to identify a convenient way to encode input data that balanced information loss with simplicity. This paper developed an Artificial Neural Network (ANN) model and applied it to the fire dataset collected from the Integrated Sensor System (ISS). Every neuron of the model will learn and hold weights that weigh information, which provides better accuracy. To mitigate loss functions and improve accuracy, various activation functions such as Sigmoid, Relu, and optimizer Stochastic Gradient Descent (SGD), Adam, and Adamax are used in the designed model. The results demonstrated that the prediction accuracy of the ANN model with Adam as the optimizer is better than that of the other two optimizers. The findings also show that the ANN model performs well in terms of prediction accuracy and is also better suited to the sensor fire dataset