Abstract:

This paper describes the principle and algorithm of the Adaboost method. By introducing the principle of free- way incident detection and parameter choice, the topology of neural networks is employed in this paper, and an improved freeway incident-detection algorithm and the steps of its processing are presented based on the Adaboost algorithm. In addi- tion, a simulation experiment is conducted to test the feasibility and validity of this algorithm.The result of the experiment shows that this algorithm can highly enhance the performance of weak classification algorithms with a higher detection rate and a lower false alarm rate, which assesses the effectiveness of its application to freeway incident detection.

Key words: Adaboost, freeway incident detecdon, neural network, class