This paper compares three approaches to multiclass learning problems: direct application of multiclass algorithms (e.g., C4.5 and CART), application of binary concept learning algorithms to learn individual binary functions for each class, and the use of distributed output representations. The authors introduce a new technique called error-correcting output codes (ECOC), which employs error-correcting codes as a distributed output representation. They demonstrate that ECOC improves the generalization performance of both C4.5 and backpropagation on a wide range of multiclass learning tasks. The approach is robust to changes in the size of the training sample, the assignment of distributed representations to classes, and the application of overfitting avoidance techniques. Additionally, ECOC can provide reliable class probability estimates. The paper concludes with a discussion of open questions, including the independence of errors in different bit positions and the relationship between ECOC and ensemble methods.This paper compares three approaches to multiclass learning problems: direct application of multiclass algorithms (e.g., C4.5 and CART), application of binary concept learning algorithms to learn individual binary functions for each class, and the use of distributed output representations. The authors introduce a new technique called error-correcting output codes (ECOC), which employs error-correcting codes as a distributed output representation. They demonstrate that ECOC improves the generalization performance of both C4.5 and backpropagation on a wide range of multiclass learning tasks. The approach is robust to changes in the size of the training sample, the assignment of distributed representations to classes, and the application of overfitting avoidance techniques. Additionally, ECOC can provide reliable class probability estimates. The paper concludes with a discussion of open questions, including the independence of errors in different bit positions and the relationship between ECOC and ensemble methods.