Deep Recurrent Neural Network (Deep-RNN) for Classification of Nonlinear Data

Debasmita Mishra,Bighnaraj Naik,Ronali Madhusmita Sahoo,Janmenjoy Nayak
DOI: https://doi.org/10.1007/978-981-15-2449-3_17
2020-01-01
Abstract:Data mining is the most challenging approach that uses the method of extracting the most interesting patterns from a large storage of database. Classification, a supervised learning method, is mostly applicable method of data mining. In this paper, we have used different classification techniques to differentiate the results for different data sets. Deep learning or hierarchical learning is the part of machine learning which mainly follows the widely used concepts of a neural network. There are many deep learning architectures such as deep neural networks, deep belief networks, recurrent neural networks, convolutional neural networks, etc. In this paper, we have used the concept of deep recurrent neural network (Deep-RNN) to train the model for a classification task. RNN follows a method for weight updation which is known as Backpropagation Through Time (BPTT) and we have used the concept of Deep-RNN by following the concepts of both forward pass and backward pass. Simulation results are quite impressive as compared to earlier developed machine learning models.
What problem does this paper attempt to address?