In this paper, we explore the performance of the quantum-inspired Quantum Extreme Learning Machine (QELM) for molecular property prediction using regression. Specifically, we investigate how quantum-inspired techniques like Tensor Networks (TNs) such as the Time-Dependent Variational Principle (TDVP) with Matrix Product States (MPS) can be utilized for the QELM algorithm. We applied the QELM technique to predict activity (ACT) values based on molecular descriptors from the Merck Molecular Activity Challenge (MMACD) dataset. Overall, the results show that TNs can balance performance and accuracy in simulating quantum dynamics, allowing the simulation of datasets with a high number of features. This study underscores the potential of tensor networks as quantum-inspired algorithms to enhance the capability of quantum machine learning algorithms to study datasets with large numbers of features.
In this paper, we explore the performance of the quantum-inspired Quantum Extreme Learning Machine (QELM) for molecular property prediction using regression. Specifically, we investigate how quantum-inspired techniques like Tensor Networks (TNs) such as the Time-Dependent Variational Principle (TDVP) with Matrix Product States (MPS) can be utilized for the QELM algorithm. We applied the QELM technique to predict activity (ACT) values based on molecular descriptors from the Merck Molecular Activity Challenge (MMACD) dataset. Overall, the results show that TNs can balance performance and accuracy in simulating quantum dynamics, allowing the simulation of datasets with a high number of features. This study underscores the potential of tensor networks as quantum-inspired algorithms to enhance the capability of quantum machine learning algorithms to study datasets with large numbers of features.
Quantum Extreme Learning Machine (QELM) is an emerging hybrid quantum machine learning framework that leverages quantum system dynamics to enhance classical models. However, QELM can suffer from the exponential concentration problem, where excessive entanglement reduces model expressivity. In this work, we gain insight into this challenge and demonstrate how tensor network methods specifically, the Time Dependent Variational Principle (TDVP) with Matrix Product States (MPS) can efficiently simulate quantum systems while controlling entanglement and mitigating exponential concentration. Using numerical experiments on the Modified National Institute of Standards and Technology (MNIST) dataset, we show that time-evolving an MPS system modeled as a chain of Rydberg atoms produces high-quality data embeddings with low classical computational overhead. Our findings indicate that exact simulation of quantum dynamics is not necessary for strong machine learning performance; even approximate quantum embeddings can yield competitive results. Furthermore, we observe that both increased disorder in the quantum state achieved by tuning Hamiltonian parameters and careful control of entanglement directly correlate with improved model accuracy, highlighting the importance of these factors in optimizing QELM performance.
Quantum machine learning (QML) has been demonstrated to have great potential in revolutionizing the process of drug discovery by dealing with the problem of high computational complexity of molecular simulations. Conventional computational techniques are not very effective in predicting the properties of molecules and pharmaceutical activity because the simulation of large molecules is too complex. This paper will discuss how QML algorithms can be used, specifically variational quantum circuits and quantum kernel methods, to improve the predictability of molecular properties and accelerate the process of drug discovery. To enhance the prediction accuracy, the research use quantum-enhanced feature extraction together with classical machine learning technologies. Research methodology is applied to datasets of molecular structure and biological activity in order to provide a more useful model to learn about how molecular features relate to biological outcomes using quantum machine learning models. These results suggest that it is 15 % more accurate in prediction and 20 % faster in computing high-dimensional molecular simulations compared to standard machine learning algorithms. The efficiency of the QML-based method is statistically proven with an accuracy 89%, precision 87%, and an F1-score of 0.91. The results point to the possible ability of QML to decrease the time and computational resources spent on the exploration of drug discovery, providing a new direction for the improvement of molecular property prediction. The study is a contribution to the emerging area of QML in pharmaceutical use and demonstrates that it can simplify drug discovery and enhance the accuracy with which predictions are made. The paper demonstrates that QML may produce a significant impact on the drug development process by simulating the molecular properties faster, more effectively, and more accurately.
Abstract Federated learning (FL) has gained significant traction across diverse industries, which allows multiple clients or institutions to enhance model performance and outcomes while preserving data privacy collaboratively. In recent years, tensor networks (TNs) have become important in machine learning because they allow the compact representation of high-dimensional tensors by decomposing them into lower-dimensional components with polynomial complexity. The application of TNs in FL is a natural extension because of its flexible framework for representing and optimizing models. Inspired by quantum computing principles, we have integrated a quantum-inspired tensor network into the FL framework. This framework focuses on a one-dimensional matrix product state (MPS) tensor network (TN) in a federated setting (FedTN), with data distributed across homogeneous and heterogeneous partitions among clients. Our experiments demonstrate that tensor network-based federated learning can be made practical, as FedTN is robust to the unbalanced and non-IID data distributions typically encountered in such settings. Our research assessed the effectiveness and feasibility of comparing quantum-inspired TN and conventional methods, evaluating their performance, and exploring the benefits of incorporating quantum principles in FL settings. Furthermore, we have investigated its performance when training for many local epochs (large E) between the averaging steps.
Machine learning has been increasingly utilized in the field of biomedical research to accelerate the drug discovery process. In recent years, the emergence of quantum computing has led to the extensive exploration of quantum machine learning algorithms. Quantum variational machine learning algorithms are currently the most prevalent, but they face issues with trainability due to vanishing gradients. An emerging alternative is the quantum reservoir computing (QRC) approach, in which the quantum algorithm does not require gradient evaluation on the quantum hardware. Motivated by the potential advantages of the QRC method, we apply it to predict the biological activity of potential drug molecules based on molecular descriptors. We observe more robust QRC performance as the size of the data set decreases, compared to standard classical models, a quality of potential interest for pharmaceutical data sets of limited size. In addition, we leverage the uniform manifold approximation and projection technique to analyze structural changes as classical features are transformed through quantum dynamics, and we find that quantum reservoir embeddings provide better separability between active and inactive compounds in the low-dimensional learned feature space.
Quantum extreme learning machines (QELMs) are unconventional computing architectures that bear remarkable promise in both classical and quantum machine-learning tasks, such as the estimation of quantum state properties. However, the probabilistic nature of quantum measurements demands extensive repetitions for training to precisely estimate expectation values, imposing stringent trade-offs among experimental resources, acquisition time, and signal-to-noise ratio, particularly for large datasets. Here, we introduce a paradigm shift by training the QELM exclusively with intense classical fields, namely twin beams generated by stimulated emission, while performing inference directly on previously unseen genuine quantum input states to predict their quantum properties. This strategy dramatically reduces acquisition times while substantially enhancing the signal-to-noise ratio. Using frequency-bin encoded biphoton states, implemented here for the first time in a quantum machine learning architecture, we demonstrate entanglement witnessing of two-qubit states with (93 +- 4)% accuracy, multi-dimensional entanglement detection, and learning of the Hamiltonian governing photon-pair generation with a fidelity of (96 +- 4)%. Our results open a new pathway toward faster and more robust training of photonic quantum extreme learning machines for quantum feature extraction.
Quantum and quantum-inspired machine learning has emerged as a promising and challenging research field due to the increased popularity of quantum computing, especially with near-term devices. Theoretical contributions point toward generative modeling as a promising direction to realize the first examples of real-world quantum advantages from these technologies. A few empirical studies also demonstrate such potential, especially when considering quantum-inspired models based on tensor networks. In this work, we apply tensor-network-based generative models to the problem of molecular discovery. In our approach, we utilize two small molecular datasets: a subset of 4989 molecules from the QM9 dataset and a small in-house dataset of 516 validated antioxidants from TotalEnergies. We compare several tensor network models against a generative adversarial network using different sample-based metrics, which reflect their learning performances on each task, and multiobjective performances using 3 relevant molecular metrics per task. We also combine the output of the models and demonstrate empirically that such a combination can be beneficial, advocating for the unification of classical and quantum(-inspired) generative learning.