SEQUENT: Towards Traceable Quantum Machine Learning using Sequential Quantum Enhanced Training

Therefore, hybrid approaches have been proposed, where the power of both classical and With classical computation evolving towards performance quantum computation are united for improved results saturation, new computing paradigms like (Bergholm et al., 2018). By this, it is possible to leverage quantum computing arise, promising superior performance the advantages of quantum computing for tasks in complex problem domains. However, current with parameter spaces that cannot be computed solely architectures merely reach numbers of 100 quantum by quantum computers due to hardware and simulation bits (qubits), prone to noise, and classical computers limitations. Within those hybrid algorithms the run out of resources simulating similar sized quantum part is, analogue to the classical deep neural systems (Preskill, 2018). Thus, most real world applications networks (DNNs), represented by so called variational are not yet feasible solely relying on quantum quantum circuits (VQCs), which are parameterized compute. Especially in the field of machine learning, and can be trained in a supervised manner where parameter spaces sized upwards of 50 million using labeled data (Cerezo et al., 2021). For hybrid are required for tasks like image classification, machine learning, we will from hereon refer to VQCs the resources of current quantum hardware or simulators as quantum parts and to DNNs as classical parts.