Figure 5
From: Estimating the phase volume fraction of multi-phase steel via unsupervised deep learning
InfoGAN and MLP classifier structures are shown inside dashed black and red boxes, respectively. In Phase I, InfoGAN is trained. In Phase II, the generator creates samples and labels \(({x}^{^{\prime}}, {c}^{^{\prime}})\) that are given as input to MLP for training. Then, the trained model performs the classification of raw data.
