Skip to main content

Table 1 Test classification accuracy of the proposed HNF for different datasets using random matrices

From: High-dimensional neural feature design for layer-wise reduction of training cost

Dataset Size of training data Size of testing data Input dimension (P) Number of classes (Q) Proposed HNF ELM Proposed HNF State-of-the-art [reference]
      Accuracy n(1) L Accuracy Accuracy n(1) L  
Letter 13,333 6667 16 26 93.3 250 5 88.3 94.6 1000 3 95.8 [25]
Shuttle 43,500 14,500 9 7 99.3 250 5 99.0 99.6 1000 3 99.9 [19]
MNIST 60,000 10,000 784 10 97.1 1000 5 96.9 97.7 4000 3 99.7 [26]