Use Cases
Quantum-safe Cryptography
- Quantum computing poses a security risk for the commonly used RSA encryption.
- New encryption schemes have been developed and are already available (see, e.g., https://openquantumsafe.org/).
- Are quantum-safe cryptography solutions ready for adoption?
- What does it need to integrate quantum-safe encryption into current systems?
- What are the implications regarding run-time, effort, and security?
Quantum Risk Analysis
- Risk analysis is concerned with the evaluation of statistical properties given some complex financial product.
- Quantum risk analysis can leverage the quantum computer’s ability to efficiently operate on probability amplitudes.
- Multiple approaches have been investigated (see, e.g., Quantum risk analysis | (nature.com)).
- How do benchmarking and scaling properties of such algorithms look like?
Quantum-enhanced Anomaly Detection
- Generative adversarial networks represent suitable methods for anomaly detection.
- One possible variation involves replacing the classical generator with a quantum neural network.
- Central motivation is Google’s quantum supremacy experiment | (nature.com): NISQ devises are good for sampling.
- But how much training is needed for such an approach?
- Can we achieve similarly good results as with classical methods?
Quantum Deep Hedging
- Quantum deep hedging is an approach to learning a hedging strategy with quantum neural networks.
- Its benefits have been shown by papers (see, e.g., Deep Hedging | (arxiv.org) or Quantum Deep Hedging | (arxiv.org)).
- These methods can be applied to arbitrary market settings and complex payoff functions.
- Both classical and quantum deep hedging outperform standard hedging approaches.
- How do the expressivities of quantum and classical deep hedging approaches compare?
Quantum Optimization
- Optimization problems are ubiquitous. Finance examples are asset allocation, optimal arbitrage opportunities, or option pricing.
- Finding a good optimizer for a given problem is a difficult task and generally involves complex modelling of the problem.
- Quantum optimizers help traverse the cost landscape more efficiently because of tunnelling effects.
- There exist promising quantum-inspired optimizers that are not restricted by nearest-neighbor and pairwise interactions (see, e.g., Simulated Quantum Annealing | (github.com)).
- Are there differences in the performance of classical, quantum and quantum-inspired optimizers?
Explainable AI for Quantum Models
- Quantum circuits are difficult to simulate at large scales, which impedes their explainability.
- These explainers still take a long time to be evaluated.
- Instead, a viable procedure could be classical surrogates for quantum models.
- A classical surrogate is a model that approximates the quantum model up to a small error.
- Fourier-based surrogates are promising and can be used as explainers (see, e.g., Classical surrogates for quantum learning models | (arxiv.org))