Latent Style-Based Quantum Wasserstein GAN for Drug Design
The paper proposes a latent style-based quantum Wasserstein GAN (QGAN) for drug design, combining a variational autoencoder that maps molecular structures into a latent space with a hybrid quantum GAN. It benchmarks against classical MOSES suite models, achieving comparable performance with far fewer trainable parameters (110 versus 700,000). Experiments include training on noiseless quantum simulators with up to 15 qubits and an inference-only demonstration on a 156‑qubit IBM Heron quantum computer using a five‑qubit setup, sampling 2,500 new molecules to study quantum noise effects.
The paper proposes a latent style-based quantum Wasserstein GAN (QGAN) for drug design, combining a variational autoencoder that maps molecular structures into a latent space with a hybrid quantum GAN. It benchmarks against classical MOSES suite models, achieving comparable performance with far fewer trainable parameters (110 versus 700,000). Experiments include training on noiseless quantum simulators with up to 15 qubits and an inference-only demonstration on a 156‑qubit IBM Heron quantum computer using a five‑qubit setup, sampling 2,500 new molecules to study quantum noise effects.
It shows that quantum‑enhanced generative models can reduce parameter counts and potentially accelerate drug discovery, supporting Swiss AI’s goal of advancing cutting‑edge AI for life‑sciences innovation.
Original source record. Open the original record to verify the underlying announcement.