Analysis of Bio-Inspired Optimization Algorithms in Feature Selection for Facial Expressions Recognition
M. Jagadeesh, B. Baranidharan · 2022 4th International Conference on Inventive Research in Computing Applications (ICIRCA) · 2022
There are several popular methods for extracting facial expression features but finding the most significant features of a particular data sample and weeding out the rest that doesn't help us make better decisions has always been difficult. The Feature selection algorithm aims to solve this issue by removing irrelevant and repeated features. Feature selection involves including vital features or removing unimportant ones without changing them. Statistics methods traditionally used are ineffective as they increase the number of data samples and features to each sample. Feature selection techniques are used for several reasons [1]: (1) It eliminates unnecessary features from the model and reduces its complexity, (2) Accelerates the training process of ML algorithms, (3) Dimensional reduction reduces over-fitting, and (4) Builds a more accurate model. Recent advancements in swarm intelligence (SI) technologies have helped overcome these challenges. This research study compares various feature selection methods and then analyzes the meta-heuristic approaches based on swarm intelligence to solve the optimization challenges.