Using hybrid quantum-classical convolutional neural networks to id
summary
Hybrid quantum-classical convolutional nets replace some convolutional layers with quantum circuits and train variational parameters, evaluated on breast ultrasound (≈600 images) for binary malignancy and 3D lung CT (compressed from 128×128×1064) for binary nodules. Two encodings compared: threshold vs higher-order (angle/entangling) encoding, and two circuits: basic entangling vs strongly entangling layers. Results: QC-CNNs beat classical CNN on the small 2D set (best with higher-order + basic entangler), while on 3D classical still leads but QC variants approach parity and benefit from stronger entanglement.
AI-written summary · about
chapters (6)
- 0:00Start
- 1:19Introduction to Our Institute
- 4:42Classical Convolutional Neural Networks
- 5:45Hybrid Quantum Classical Conditional Networks
- 8:59Threshold Encoding
- 14:05Summary
related talks
26mBeyond the Hype: Hybrid Quantum-Classical AI Agents for Scientific Discovery
22mSecurity of Quantum Machine Learning Systems
15mEnsuring Robustness in Hybrid Quantum Neural Networks
44mQuantum machine learning: Prospects and challenges
32mQuantum Artificial Intelligence
34mDisease Detection in the Age of Quantum Computing
17mQuantum and Quantum-Inspired AI Solutions: Bridging Innovation and Industry Readiness
34mGraph Neural Networks for Classical and Quantum Channel Coding