Random Feature Pruning for Multi-Class Classification
This work addresses redundancy in features extracted by Weighted sums of random kitchen sinks (W-RKS) for multi-class classification. The authors introduced the Successive Empirical Excess Risk Difference Maximization (SEED-MAX) algorithm to recursively select the most informative, parsimonious feature set. Combined with a Fisher test, this method achieved about 92% feature pruning while maintaining high accuracy, proving its effectiveness in complex classification tasks.