Why Error Correcting Output Coding Works

Previous research has shown that a technique called error correcting output coding ECOC can dramatically improve the classi cation accuracy of supervised learning algorithms that learn to classify data points into one of k classes This paper presents an empirical investigation of why the ECOC technique works particularly when employed with decision tree learning methods It concludes that an important factor in the success of the method is the nearly random behavior of decision tree algorithms near the root of the decision tree when applied to learn di cult decision boundaries The results also show that deliberately injecting randomness into decision tree algorithms can signi cantly improve the accuracy of voting methods that combine the guesses of multiple decision trees to make classi cation decisions