GED‐CRN Breaks the Data Barrier: High‐Fidelity Electron Density Prediction Using Only 19 Training Molecules

We present GED‐CRN, a 3D convolutional residual network that achieves quantum‐chemical accuracy (MAE ) in predicting electron densities for AIE‐active systems using only 19 training molecules—overcoming the data scarcity bottleneck via spatial cube sampling ( ) of pro‐molecular densities and nuclear potentials. The model demonstrates faster computation than MP2 while preserving AIE‐critical features ( Å vdW surface error), enabling high‐throughput screening of ‐conjugated materials with 50% lower error than conventional organic systems, as validated on QM9 and ASBase datasets. This few‐shot learning paradigm bridges data‐efficient quantum ML with functional luminescent material design.