Convolutional Neural Network &Keras and Compared with Random Forest for Traffic Sign Recognition with Improved Accuracy

Abstract

Aim: To predict Accuracy of Uveitis symptoms using blurred vision by comparing logitboost and random committee. Materials and methods: Logitboost with a sample size of 10 (N=10) and Random committee with a sample size of 10 (N=10) were iterated numerous times for predicting uveitis symptoms accuracy percentage. The required samples for this analysis are calculated using G power calculation. The minimum power for analysis is set at 0.8, while the maximum allowed error is set to 0.2. Result: Results proved that the Random committee got significant results with 76% accuracy compared to Logitboost with 72% accuracy. The statistical significance difference p = 0.775 (p<0.005) Free example T-test value states that the results in the study are insignificant. Conclusion: Random committee appears to be performing better when compared to Logitboost for predicting Uveitis symptoms.