Multiclass Classification Based on Meta Probability Codes

This paper proposes a new approach to improve multiclass classi ̄cation performance by employing Stacked Generalization structure and One-Against-One decomposition strategy. The proposed approach encodes the outputs of all pairwise classi ̄ers by implicitly embedding twoclass discriminative information in a probabilistic manner. The encoded outputs, called Meta Probability Codes (MPCs), are interpreted as the projections of the original features. It is observed that MPC, compared to the original features, has more appropriate features for clustering. Based on MPC, we introduce a cluster-based multiclass classi ̄cation algorithm, called MPC-Clustering. The MPC-Clustering algorithm uses the proposed approach to project an original feature space to MPC, and then it employs a clustering scheme to cluster MPCs. Subsequently, it trains individual multiclass classi ̄ers on the produced clusters to complete the procedure of multiclass classi ̄er induction. The performance of the proposed algorithm is extensively evaluated on 20 datasets from the UCI machine learning database repository. The results imply that MPC-Clustering is quite e±cient with an improvement of 2.4% overall classi ̄cation rate compared to the state-of-the-art multiclass classi ̄ers.