Gender Classification Using K-Nearest Neighbors and Logistic Regression Models
Vijay Madaan, Neha Vaishnavi Sharma · 2024
This work provides a hybrid model for gender identification integrating face features with logistic regression with K-Nearest Neighbors. The model reached an amazing accuracy of 96% with a balanced loss and good performance throughout multiple classes. Especially in hyperparameter tuning, overfitting, and resolution changes were seen during the development. The first training phase concentrated on problems achieving optimal convergence, which led to necessary modifications in the training calendar. Using architectural and training approaches creation of the model, these issues were solved, so producing a model consistent and accurate for gender classification tasks. The excellent low loss of the earlier model reveals its resistance and longevity by outstanding precision.