WiMi Unveils Quantum CNN with 6% Accuracy Boost and 30% Fewer Parameters
WiMi unveiled a Multi-Scale Fusion Quantum Deep Convolutional Neural Network that delivers over 6% accuracy gains on text classification benchmarks while reducing parameters by more than 30% versus classical CNNs. The model outperforms QRNN, QSAM and QTF by 4%–10% and maintains stable performance in noisy quantum hardware simulations.
1. Breakthrough QDCNN Architecture
WiMi introduced a Multi-Scale Fusion Quantum Deep Convolutional Neural Network that features quantum depthwise separable convolution and a multi-scale feature fusion mechanism to unify word-level and sentence-level modeling within a single quantum framework.
2. Performance Improvements
Benchmark results show over 6% accuracy gains on standard text classification datasets with more than 30% fewer parameters than classical CNNs; the model outperforms QRNN, QSAM and QTF by 4%–10% and retains stability under noisy quantum hardware simulations.
3. Industry Implications
This launch marks a key advance toward practical quantum natural language processing, demonstrating scalable, low-energy quantum architectures that could drive competitive differentiation and future revenue opportunities for WiMi Hologram Cloud.