Fig. 5: Color-mixed pattern classification using a 3D FeNAND-based neural network.
a Schematic illustration of color classification using 3D FeNAND and CMOS neurons. The images fabricated by randomly adding the box and line patterns with red, green, or blue colors were used as test and training images. The patterns were passed through color filters (i.e., red, green, and blue), and the filtered value was converted as the input voltage to the BLs. The output ISL was measured and used as the input for the neurons (op-amp). b Example of color-mixed pattern classification using 3D FeNAND-based neural network. The RL, RB, GL, GB, BL, and BB stands for the red line, red box, green line, green box, blue line, and blue box, respectively. For the mixed pattern containing the red line, green box, and blue box, only the corresponding output neurons showed high neuron outputs. c Classification result for 20 input patterns. The summed neuron output showed high output only for the correct label.
