Enhancing Accuracy in Sign Language Recognition Using Linear Regression Algorithm with Gradientdescent Over K Nearest Neighbors Algorithm

Sign language recognition is an essential task in closing the communication gap between hearing-impaired people and the rest of the population. This paper suggests a better machine learning method based on the Linear Regression algorithm with Gradient Descent optimization to enhance the accuracy of sign language recognition. The performance of this method is compared with the conventional K Nearest Neighbors (KNN) algorithm using a hand gesture feature dataset obtained from video frames. Accuracy and model efficiency are measured using evaluation metrics like Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and R2 score. The experimental findings show that Linear Regression with Gradient Descent performs much better than KNN by more accurately capturing continuous patterns of hand motion and minimizing prediction errors. This research points to the promise of sophisticated regression methods in improving sign language recognition systems and justifies the creation of intelligent assistive technologies for real-time, accurate sign language interpretation.