Quantum Machine Learning

Expected Outcome:

• Integration of quantum computing into data pre-processing pipelines and learning workflows for data-heavy or computationally intensive tasks, demonstrating clear improvements in processing speed, computational complexity, modelling accuracy, and reduced sample requirements at scales achievable with NISQ-era devices,

• Reliable and scalable Quantum Machine Learning (QML) models and algorithms, integrated with existing AI frameworks and pipelines, enabling faster data processing, improved prediction accuracy, and enhanced computational capabilities,

• Validated quantum-enhanced AI methods demonstrating measurable improvements over classical baselines in terms of speed, accuracy, data efficiency, complexity, or scalability, supported by rigorous benchmarking and complexity analysis,

• Robust, noise-aware QML techniques suitable for NISQ hardware , including error-mitigation strategies and algorithmic adaptations that improve reliability, performance, and reproducibility on real quantum processors,

• Demonstrators or proof-of-concept applications showcasing the relevance of QML for real-world challenges (e.g. climate and environmental modelling, Earth observation, healthcare and life sciences, materials discovery, finance, robotics, manufacturing, and cybersecurity),

• Strengthened European leadership and technological sovereignty in quantum computing and trustworthy AI, supported by cross-sector collaboration, knowledge transfer, and contributions to emerging standards, benchmarks and best practices.

• Enhanced collaboration across quantum computing, machine learning and application domains, fostering a coordinated European QML research and innovation community.
Scope:
Proposals are expected to address multiple key research directions in Quantum Machine Learning (QML), targeting both scientific excellence and industrial relevance. Proposals should clearly outline how to contribute to the development, validation and demonstration of quantum-enhanced AI approaches, with clear pathways towards practical applications. The proposed work should strengthen Europe’s scientific and technological capabilities in quantum computing and accelerate the industrial uptake of quantum-enhanced AI solutions.
Activities may include, but are not limited to

• design and analysis of quantum, quantum-inspired or hybrid QML algorithms,

• performance modelling, complexity analysis and benchmarking of quantum-enhanced AI methods,

• development of error-mitigation and noise-aware strategies tailored to QML workloads,
Proposals should advance scalable QML algorithms capable of addressing large-scale, computationally intensive problems, this includes approaches that

• can manage massive data volumes and complex computational tasks,

• enable faster data processing and improved predictive performance in relevant application domains (e.g. hydrologic research, climate modelling, terrain classification from satellite remote sensing, drug discovery, and image-based medical diagnosis).
Many current QML methods remain closely inspired by classical algorithms and therefore do not yet achieve genuine quantum advantage, requiring further developments in

• quantum-native learning models,

• efficient quantum kernels and quantum feature mappings,

• algorithms demonstrating provable or empirical advantages over classical approaches.
Proposals may also include formal complexity analyses, identification of problem classes that can benefit from quantum acceleration.
Developments should address multiple of the following key research directions :
Quantum Supervised Learning (QSL) :
Quantum Supervised Learning investigates how quantum algorithms can accelerate or improve the training of supervised learning models, offering novel opportunities to explore quantum–classical learning theories and enabling industry to shorten development cycles and enhance performance in data-intensive domains such as finance, healthcare, and Earth observation. Integrating QSL into existing pipelines may help overcome computational bottlenecks and enable more efficient processing of high-dimensional data

• Proposals should explore quantum algorithms and quantum subroutines that can be integrated into classical AI pipelines to mitigate computational bottlenecks, improve efficiency in high-dimensional or large-scale settings, and deliver measurable performance gains. Activities should include the development and validation of such methods, with demonstrations in relevant academic or industrial use cases.
Quantum Convolutional Neural Networks (QCNNs)
Quantum Convolutional Neural Networks combine the conceptual strengths of classical convolutional architectures with the computational advantages of quantum processing, offering a promising testbed for exploring expressivity and efficiency in hybrid models and providing industry with a practical pathway to near-term quantum advantage by using QPUs during training while retaining classical inference for scalability and compatibility with existing AI systems

• Proposals should explore development and evaluation of hybrid QCNN architectures that leverage quantum processing for training, investigate their expressivity, performance and robustness, and demonstrate their applicability to industrial challenges such as pattern recognition, vision-based analytics or complex classification tasks.
Learning with Quantum Models
Quantum models integrate quantum computation directly into both training and inference through parametrised quantum circuits and quantum kernels, enabling advances in understanding quantum data representations, feature expressivity and generalisation theory, while offering industry benefits such as reduced model complexity, faster training cycles and lower data requirements - particularly in data-scarce or high-value domains (e.g. drug discovery, autonomous systems, materials engineering)

• Proposals should explore the design and evaluation of quantum-native learning…