Neural network and deep learning techniques are essential tools for data scientists when analyzing big data for forecasting and classification. In supervised learning, data sets are divided into training sets and test sets, and neural network repeatedly adjusts the weight of data to better represent the actual data. This book offers a practical guide to performing a neural network experiment with RapidMiner, which readers can follow step-by-step. For big data, especially non-linear data, deep learning can be employed. This chapter introduces two types of deep learning: convolutional neural networks (CNN) for picture analysis and recurrent neural networks (RNN) for sequential or time series data. The book provides a demonstration of both techniques using RapidMiner, making it accessible to readers who wish to deepen their understanding of these powerful tools.
TopIntroduction
In the early days when people first mentioned artificial intelligence, there was a lot of motivation to develop artificial intelligence capable of human-like responses. However, the challenge for the era was that artificial intelligence computing required high-performance computing technology and large volumes of experimental data, resulting in the gradual advancement of artificial intelligence. Now in the era of high-performance computing technologies such as GPUs with a high-speed network (Telikani, Shahbahrami, & Gandomi, 2021), data scientists are able to use data mining techniques to operate on high-performance computing technologies and collect big data for experimentation using techniques that are suitable for such processing. Deep learning technique is used to analyze unstructured data consisting of images, sounds, and text (Fernando et al., 2021: Hongyi Zhu, Samtani, Brown, & Hsinchun Chen, 2021). There are two types of deep learning: Convolutional Neural Networks and Recurrent Neural Networks (Lakshmi Devi & Samundeeswari V, 2021: Snineh et al., 2021). The deep learning model modulates the pre-configuration of data for processing in conjunction with neural network techniques. Artificial Neural Networks (ANNs) is a data mining technique that offers both classification and numerical predictions, which is considered supervised learning (Thankachan, Prakash & Jothi, 2021). Data scientists must first teach machines to learn before testing the data. In the case data scientists classify data with other data mining techniques, the data are classified by a linear plane. On the contrary, using neural networks, data scientists can classify data that are closely attached to facts with a Non-Linear Function. The neural network is, therefore, used in 2 ways: first, Pattern Recognition, such as user face recognition for identification or authentication for accessing the phone (Ghorpade & Koneru, 2021), and finally Forecasting such as forecasting the trend of stock prices (Chinnarasri, Nonsawang & Supharatid, 2012).
This book discusses the principles of artificial neural network and deep learning in both CNN and RNN formats as follows: