Study of Different Regression Methods, Models and Application in Deep Learning Paradigm
Arpita Shome, Gunjan Mukherjee, Arpitam Chatterjee, Bipan Tudu · Auerbach Publications eBooks · 2024
Deep learning relies heavily on regression techniques and models because they make it possible to predict continuous dependent variables from a set of input characteristics. As predictive model, the regression has been widely used in FNN (Feed forward Neural Networks), CNN (Convolutional Neural Networks), LSTM (Long Short-Term Memory), RNN (Recurrent Neural Networks), GRUs (Gated Recurrent Units) and so forth. The particulars of the challenge, such as the type of input data, the existence of time relationships, or the requirement to capture geographical aspects, determine the type of model approach to be used. The best approach for a particular regression problem must be determined through experimentation and model review. The advent of deep concept has revolutionized the industrial productivity and impacted many allied research fields. Some of the tasks include determination of a person’s age from their facial appearances, projecting the value of stocks based on previous data, and predicting housing prices based on characteristics like location and size, and so forth. Deep learning models are made up of interrelated layers of artificial neurons and are capable of capturing the intricate nonlinear interactions between input variable sets and the desired outcomes, making precise predictions through the back propagation approach which also leads to a reduction of the gap between anticipated and desired outcomes. The proposed review paper mainly highlights different regression models in a myriad of application scopes along with different techniques in the deep learning domain and its further modification with the increasing complexity of diversified problems.