A Novel Parameter Initialization Technique Using RBM-NN for Human Action Recognition

<div>(a) Mean squared error (MSE) loss over training epochs; (b) mean squared logarithmic error (MSLE) loss over training epochs; (c) mean absolute error (MAE) loss over training epochs.</div>

Computational Intelligence and Neuroscience

fig30

Figure 30

Figure 30: A Novel Parameter Initialization Technique Using RBM-NN for Human Action Recognition 