Insider Brief
- Ludwig Maximilian University of Munich-led team of researchers suggest that quantum neural networks (QNNs) could be a tool for chemical and pharmaceutical companies to test surrogate functions.
- Surrogate functions are simplified functions used to model processes in tasks, such as drug interactions.
- The team’s findings suggest QNNs would outperform classical neural networks.
A major challenge in the chemical and pharmaceutical industries is finding a cost-effective way to model and evaluate surrogate functions that closely approximate complex black box functions. In the pharmaceutical industry, surrogate functions — which are simplified functions used to model processes — are often relied on in tasks, such as drug interactions, to optimize and streamline development efforts.
The conventional methods of classical machine learning often struggle to accurately solve this problem due to the limited and noisy datasets commonly encountered in practical applications. As a result, chemical companies worldwide are keenly interested in exploring novel approaches to address this issue.
In a study, a Ludwig Maximilian University of Munich-led team of researchers suggest that quantum neural networks (QNNs) offer a highly promising solution. These networks, based on principles of quantum mechanics, show the potential to surpass their classical counterparts when trained on small and noisy datasets.
The researchers conclude in what they claim is the first practical exploration of utilizing QNNs as surrogate models for higher-dimensional and realistic data: “In this paper, we have demonstrated that Quantum Surrogate Models based on QNNs can offer an advantage over classical ANNs in terms of prediction accuracy for substantially more difficult datasets than those used in literature previously, when the sample size is scarce and substantial noise is present.”
Through extensive experimentation, the team reports its QNN outperformed a minimalist classical artificial neural network when dealing with noisy and limited datasets. These results provide empirical evidence supporting the advantages of quantum surrogate models.
The researchers also measured the performance of current NISQ (Noisy Intermediate-Scale Quantum) hardware in the experiments and estimated the necessary gate fidelities required to replicate the simulation results, suggesting the practical feasibility of implementing QNNs in real-world scenarios.
The implications of the findings are significant for the chemical and pharmaceutical industries, according to the team. By leveraging QNNs as surrogate models, companies can enhance their ability to approximate complex functions accurately, even with limited and noisy data. This has the potential to streamline the development process, leading to more efficient and cost-effective solutions.
There are limitations, the team points out. While QNNs hold promise, the field of quantum computing is still in its early stages. The hardware and gate fidelities necessary for large-scale implementation are yet to be fully realized.
The team posted their findings on the pre-print server ArXiv. Pre-print servers allow researchers to share their findings but it is not technically a peer-reviewed study.
Team members include: Jonas Stein, Michael Poppel, Philip Adamczyk, Ramona Fabry, Zixin Wu, Michael Kolle Jonas Nußlein, Danielle Schuman, Philipp Altmann and Claudia Linnhoff-Popien, all of LMU Munich; Thomas Ehmer, of Merck KGaA and Vijay Narasimhan, of EMD Electronics .
For more market insights, check out our latest quantum computing news here.
