Quantum neural networks
Talkat Near-Term Quantum Computers: Fault Tolerance + Benchmarking + Quantum Advantage + Quantum Algorithms · 2024-04-22..04-26·
summary
Presents a concrete architecture called dissipative quantum neural networks that feeds forward by applying layer-to-layer unitaries and tracing out earlier layers, uses fidelity cost to train, and derives an explicit update rule (a commutator of forward input and backward target states). Classical simulation: two-qubit problem with 20 pairs (10 training, 10 validation) reaches fidelity ~1 after ~1,000 epochs. Runs on IBM Quantum Cloud show training but noise prevents perfect fidelity; also compares favorably to QAOA and motivates a quantum no-free-lunch theorem.
AI-written summary · about
chapters (15)
- 0:00Introduction
- 0:26Machine learning overview
- 1:53Classical neural networks
- 3:43Supervised learning example
- 5:22Quantum machine learning
- 7:00Quantum network structures
- 8:46Deep quantum networks
- 10:12Feedforward process
- 12:23Loss functions and training
- 14:23Update rules and commutator
- 18:04Training and validation
- 19:04Example: Learning a unitary
- 22:33Network applications
- 25:32Future research directions
- 27:22Q&A session
related talks
44mQuantum machine learning: Prospects and challenges
15mEnsuring Robustness in Hybrid Quantum Neural Networks
17mQuantum and Quantum-Inspired AI Solutions: Bridging Innovation and Industry Readiness
32mQuantum Artificial Intelligence
27mQuantum Machine Learning With TensorFlow Quantum
28mQuantum machine learning with superconducting circuits
01h 01mTensor Networks and Quantum Algorithms
46mQuantum Machine Learning 101: An Introduction for Beginners with...