Quantum kernels outperform classical methods on engineered tasks using 15-45 qubits on IQM hardware, with detailed analysis showing error mitigation can revive advantages lost to noise, though classical simulation remains a key challenge.
#QuantumMachineLearning #QuantumAdvantage #Research
#QuantumMachineLearning #QuantumAdvantage #Research
Engineered Learning Problems on Noisy Quantum Processors
arxiv.org
November 12, 2026 at 4:37 PM
Quantum kernels outperform classical methods on engineered tasks using 15-45 qubits on IQM hardware, with detailed analysis showing error mitigation can revive advantages lost to noise, though classical simulation remains a key challenge.
#QuantumMachineLearning #QuantumAdvantage #Research
#QuantumMachineLearning #QuantumAdvantage #Research
APGEM uses contextual bandits to adaptively select error mitigation techniques during quantum machine learning training, achieving 95.5% of oracle performance across five QML paradigms while intelligently adapting to changing noise conditions.
#QuantumErrorMitigation #QuantumMachineLearning #NISQ
#QuantumErrorMitigation #QuantumMachineLearning #NISQ
Adaptive Policy-Guided Error Mitigation: Learning-Based Framework for Noise Resilience in Quantum Machine Learning
iq.fp2.dev
November 11, 2026 at 8:52 AM
APGEM uses contextual bandits to adaptively select error mitigation techniques during quantum machine learning training, achieving 95.5% of oracle performance across five QML paradigms while intelligently adapting to changing noise conditions.
#QuantumErrorMitigation #QuantumMachineLearning #NISQ
#QuantumErrorMitigation #QuantumMachineLearning #NISQ
Hybrid quantum-classical system (HQNN) detects trade-based fraud at 57.9% F1 with 5 qubits on IBM hardware—competitive with classical ML, no quantum advantage yet, but demonstrates practical quantum-hybrid viability for financial crime detection.
#QuantumMachineLearning #FinancialCrime #Research
#QuantumMachineLearning #FinancialCrime #Research
Hybrid Quantum-Classical Algorithms for Trade Surveillance
arxiv.org
November 12, 2026 at 4:32 PM
Hybrid quantum-classical system (HQNN) detects trade-based fraud at 57.9% F1 with 5 qubits on IBM hardware—competitive with classical ML, no quantum advantage yet, but demonstrates practical quantum-hybrid viability for financial crime detection.
#QuantumMachineLearning #FinancialCrime #Research
#QuantumMachineLearning #FinancialCrime #Research
Last week I had the pleasure to co-arrange the QSIP - Quantum Sweden Innovation Platform workshop hackathon on quantum machine learning, run with #IQM Quantum Computers and #QuTwo.
Materials for self study: github.com/erikkallman/...
Papers incoming.
#QuantumMachineLearning
Materials for self study: github.com/erikkallman/...
Papers incoming.
#QuantumMachineLearning
GitHub - erikkallman/QML-selfstudy
Contribute to erikkallman/QML-selfstudy development by creating an account on GitHub.
github.com
November 11, 2026 at 2:10 PM
Last week I had the pleasure to co-arrange the QSIP - Quantum Sweden Innovation Platform workshop hackathon on quantum machine learning, run with #IQM Quantum Computers and #QuTwo.
Materials for self study: github.com/erikkallman/...
Papers incoming.
#QuantumMachineLearning
Materials for self study: github.com/erikkallman/...
Papers incoming.
#QuantumMachineLearning
Researchers developed a 16-qubit conditional IQP Born machine that accurately predicts magnetization transitions at unseen temperatures, outperforming diagonal models with 6x lower error while scaling to 2048 qubits without quantum hardware resources.
#QuantumMachineLearning #IQPCircuits #News
#QuantumMachineLearning #IQPCircuits #News
Classical Training of Conditional IQP Born Machines for Quantum Simulation
iq.fp2.dev
November 7, 2026 at 10:47 PM
Researchers developed a 16-qubit conditional IQP Born machine that accurately predicts magnetization transitions at unseen temperatures, outperforming diagonal models with 6x lower error while scaling to 2048 qubits without quantum hardware resources.
#QuantumMachineLearning #IQPCircuits #News
#QuantumMachineLearning #IQPCircuits #News
Aqarios leads IgnisQNC initiative to develop quantum neural computing that transforms noise into a computational asset for practical machine learning, with built-in explainability for EU AI Act compliance.
#QuantumMachineLearning #QuantumNeuralComputing #News
#QuantumMachineLearning #QuantumNeuralComputing #News
IgnisQNC Project Harnesses Quantum Noise for Explainable Machine Learning
iq.fp2.dev
November 5, 2026 at 3:35 PM
Aqarios leads IgnisQNC initiative to develop quantum neural computing that transforms noise into a computational asset for practical machine learning, with built-in explainability for EU AI Act compliance.
#QuantumMachineLearning #QuantumNeuralComputing #News
#QuantumMachineLearning #QuantumNeuralComputing #News
The FCA establishes its inaugural Quantum Experimentation Group to explore quantum machine learning applications for detecting fraud and money laundering, partnering with financial services firms, quantum providers, and research labs.
#QuantumMachineLearning #FinancialRegulation #News
#QuantumMachineLearning #FinancialRegulation #News
FCA Launches First Quantum Experimentation Group for Financial Crime Detection
thequantuminsider.com
November 5, 2026 at 2:24 PM
The FCA establishes its inaugural Quantum Experimentation Group to explore quantum machine learning applications for detecting fraud and money laundering, partnering with financial services firms, quantum providers, and research labs.
#QuantumMachineLearning #FinancialRegulation #News
#QuantumMachineLearning #FinancialRegulation #News
Hybrid quantum-classical malware detector achieves 97.2% accuracy but reveals vulnerabilities to adversarial attacks. Dynamic training reduces feature-space attacks to 5%, though compound attacks persist at 34% success rate.
#QuantumMachineLearning #Cybersecurity #Research
#QuantumMachineLearning #Cybersecurity #Research
Hybrid Quantum-Classical Malware Detector with Adversarial Robustness Analysis
iq.fp2.dev
October 31, 2026 at 10:37 PM
Hybrid quantum-classical malware detector achieves 97.2% accuracy but reveals vulnerabilities to adversarial attacks. Dynamic training reduces feature-space attacks to 5%, though compound attacks persist at 34% success rate.
#QuantumMachineLearning #Cybersecurity #Research
#QuantumMachineLearning #Cybersecurity #Research
Symmetry reorganizes quantum Fourier model frequency spectra into orbits, mitigating vanishing expressivity. New randomized encoding enables invariant quantum models without additional qubits or circuit depth.
#QuantumMachineLearning #QuantumAlgorithms #Research
#QuantumMachineLearning #QuantumAlgorithms #Research
Fourier Symmetrization for Geometric Quantum Machine Learning
iq.fp2.dev
November 1, 2026 at 5:03 PM
Symmetry reorganizes quantum Fourier model frequency spectra into orbits, mitigating vanishing expressivity. New randomized encoding enables invariant quantum models without additional qubits or circuit depth.
#QuantumMachineLearning #QuantumAlgorithms #Research
#QuantumMachineLearning #QuantumAlgorithms #Research
IBM researchers trained neural networks to predict high-quality QAOA angles, replacing expensive per-instance classical optimization with single forward passes—achieving solver-level quality while reducing computation by orders of magnitude.
#QuantumOptimization #QuantumMachineLearning #Research
#QuantumOptimization #QuantumMachineLearning #Research
Deep Learning for Predicting Quantum Approximate Optimization Algorithm Angles
iq.fp2.dev
November 1, 2026 at 8:56 AM
IBM researchers trained neural networks to predict high-quality QAOA angles, replacing expensive per-instance classical optimization with single forward passes—achieving solver-level quality while reducing computation by orders of magnitude.
#QuantumOptimization #QuantumMachineLearning #Research
#QuantumOptimization #QuantumMachineLearning #Research
Hybrid quantum-classical malware detector achieves 97.2% accuracy but reveals vulnerabilities to adversarial attacks. Dynamic training reduces feature-space attacks to 5%, though compound attacks persist at 34% success rate.
#QuantumMachineLearning #Cybersecurity #Research
#QuantumMachineLearning #Cybersecurity #Research
Hybrid Quantum-Classical Malware Detector with Adversarial Robustness Analysis
iq.fp2.dev
October 31, 2026 at 10:37 PM
Hybrid quantum-classical malware detector achieves 97.2% accuracy but reveals vulnerabilities to adversarial attacks. Dynamic training reduces feature-space attacks to 5%, though compound attacks persist at 34% success rate.
#QuantumMachineLearning #Cybersecurity #Research
#QuantumMachineLearning #Cybersecurity #Research
Researchers developed a machine learning framework for quantum state classification with polynomial measurement scaling, enabling efficient classification of non-Gaussian states on a programmable photonic processor instead of exponential scaling.
#QuantumMachineLearning #PhotonicComputing #News
#QuantumMachineLearning #PhotonicComputing #News
Polynomial-Scaling Quantum State Classification Framework Using Photonic Processors
iq.fp2.dev
October 31, 2026 at 5:33 PM
Researchers developed a machine learning framework for quantum state classification with polynomial measurement scaling, enabling efficient classification of non-Gaussian states on a programmable photonic processor instead of exponential scaling.
#QuantumMachineLearning #PhotonicComputing #News
#QuantumMachineLearning #PhotonicComputing #News
Researchers developed a machine learning framework for quantum state classification with polynomial measurement scaling, enabling efficient classification of non-Gaussian states on a programmable photonic processor instead of exponential scaling.
#QuantumMachineLearning #PhotonicComputing #News
#QuantumMachineLearning #PhotonicComputing #News
Polynomial-Scaling Quantum State Classification Framework Using Photonic Processors
iq.fp2.dev
October 31, 2026 at 5:33 PM
Researchers developed a machine learning framework for quantum state classification with polynomial measurement scaling, enabling efficient classification of non-Gaussian states on a programmable photonic processor instead of exponential scaling.
#QuantumMachineLearning #PhotonicComputing #News
#QuantumMachineLearning #PhotonicComputing #News
HQCS enables quantum generation of arbitrary-length time series using fixed qubits, proves inverse-polynomial loss variance (avoiding exponential concentration), and demonstrates >10x better MMD than classical benchmarks on VIX financial data.
#QuantumMachineLearning #QuantumAlgorithms #Research
#QuantumMachineLearning #QuantumAlgorithms #Research
Generative Learning of Time Series with Holographic Quantum Circuit Sampling
arxiv.org
October 31, 2026 at 5:06 AM
HQCS enables quantum generation of arbitrary-length time series using fixed qubits, proves inverse-polynomial loss variance (avoiding exponential concentration), and demonstrates >10x better MMD than classical benchmarks on VIX financial data.
#QuantumMachineLearning #QuantumAlgorithms #Research
#QuantumMachineLearning #QuantumAlgorithms #Research
HQCS enables quantum generation of arbitrary-length time series using fixed qubits, proves inverse-polynomial loss variance (avoiding exponential concentration), and demonstrates >10x better MMD than classical benchmarks on VIX financial data.
#QuantumMachineLearning #QuantumAlgorithms #Research
#QuantumMachineLearning #QuantumAlgorithms #Research
Generative Learning of Time Series with Holographic Quantum Circuit Sampling
arxiv.org
October 31, 2026 at 5:06 AM
HQCS enables quantum generation of arbitrary-length time series using fixed qubits, proves inverse-polynomial loss variance (avoiding exponential concentration), and demonstrates >10x better MMD than classical benchmarks on VIX financial data.
#QuantumMachineLearning #QuantumAlgorithms #Research
#QuantumMachineLearning #QuantumAlgorithms #Research
Quantum data re-uploading architecture demonstrates universal function approximation using just one ancilla qubit with quantum inputs, enabling space-efficient quantum machine learning models that process quantum data directly.
#QuantumMachineLearning #QuantumAlgorithms #News
#QuantumMachineLearning #QuantumAlgorithms #News
Re-uploading Quantum Data: Universal Function Approximation for Quantum Inputs
iq.fp2.dev
October 29, 2026 at 1:47 PM
Quantum data re-uploading architecture demonstrates universal function approximation using just one ancilla qubit with quantum inputs, enabling space-efficient quantum machine learning models that process quantum data directly.
#QuantumMachineLearning #QuantumAlgorithms #News
#QuantumMachineLearning #QuantumAlgorithms #News
Quantum data re-uploading architecture demonstrates universal function approximation using just one ancilla qubit with quantum inputs, enabling space-efficient quantum machine learning models that process quantum data directly.
#QuantumMachineLearning #QuantumAlgorithms #News
#QuantumMachineLearning #QuantumAlgorithms #News
Re-uploading Quantum Data: Universal Function Approximation for Quantum Inputs
iq.fp2.dev
October 29, 2026 at 1:47 PM
Quantum data re-uploading architecture demonstrates universal function approximation using just one ancilla qubit with quantum inputs, enabling space-efficient quantum machine learning models that process quantum data directly.
#QuantumMachineLearning #QuantumAlgorithms #News
#QuantumMachineLearning #QuantumAlgorithms #News
EP training extends to driven-dissipative quantum systems via hidden time-reversal symmetry, enabling efficient parameter optimization without ground state preparation—validated on Ising and bosonic lattice models.
#QuantumMachineLearning #QuantumTraining #Research
#QuantumMachineLearning #QuantumTraining #Research
Equilibrium Propagation for Training Non-Equilibrium Quantum Systems
arxiv.org
October 27, 2026 at 6:47 AM
EP training extends to driven-dissipative quantum systems via hidden time-reversal symmetry, enabling efficient parameter optimization without ground state preparation—validated on Ising and bosonic lattice models.
#QuantumMachineLearning #QuantumTraining #Research
#QuantumMachineLearning #QuantumTraining #Research
Novel quantum ML framework encodes training data into parameterized Hamiltonians, providing provable generalization bounds and practical implementations on IBM Heron processors, combining theory with experimental validation.
#QuantumMachineLearning #QuantumAlgorithms #Research
#QuantumMachineLearning #QuantumAlgorithms #Research
Supervised Learning with Quantum Hamiltonian Embeddings
arxiv.org
October 20, 2026 at 4:59 AM
Novel quantum ML framework encodes training data into parameterized Hamiltonians, providing provable generalization bounds and practical implementations on IBM Heron processors, combining theory with experimental validation.
#QuantumMachineLearning #QuantumAlgorithms #Research
#QuantumMachineLearning #QuantumAlgorithms #Research
Quantum machine learning models for neuropsychiatric disorder classification performed comparably to classical models on matched features across four public cohorts using…
#QuantumMachineLearning #Neuropsychiatry #MachineLearning #FunctionalConnectivity
https://arxiv.org/abs/2610.21344
#QuantumMachineLearning #Neuropsychiatry #MachineLearning #FunctionalConnectivity
https://arxiv.org/abs/2610.21344
October 17, 2026 at 2:01 PM
Quantum machine learning models for neuropsychiatric disorder classification performed comparably to classical models on matched features across four public cohorts using…
#QuantumMachineLearning #Neuropsychiatry #MachineLearning #FunctionalConnectivity
https://arxiv.org/abs/2610.21344
#QuantumMachineLearning #Neuropsychiatry #MachineLearning #FunctionalConnectivity
https://arxiv.org/abs/2610.21344
A new arXiv paper proposes fusing classical and quantum representations in multi-view learning, aiming to improve prediction when both views capture complementary signals. The work highlights potential relevance…
#QuantumMachineLearning #AI #Cybersecurity #HybridAI
https://arxiv.org/abs/2610.21395
#QuantumMachineLearning #AI #Cybersecurity #HybridAI
https://arxiv.org/abs/2610.21395
October 17, 2026 at 12:01 AM
A new arXiv paper proposes fusing classical and quantum representations in multi-view learning, aiming to improve prediction when both views capture complementary signals. The work highlights potential relevance…
#QuantumMachineLearning #AI #Cybersecurity #HybridAI
https://arxiv.org/abs/2610.21395
#QuantumMachineLearning #AI #Cybersecurity #HybridAI
https://arxiv.org/abs/2610.21395
Quantum support vector machines demonstrated improved scaling with training data compared to variational quantum classifiers for Higgs classification, though both remained below top classical methods in this rigorous benchmark study.
#QuantumMachineLearning #HighEnergyPhysics #Research
#QuantumMachineLearning #HighEnergyPhysics #Research
Quantum Machine Learning Scalability for Higgs Boson Classification
arxiv.org
October 14, 2026 at 11:40 AM
Quantum support vector machines demonstrated improved scaling with training data compared to variational quantum classifiers for Higgs classification, though both remained below top classical methods in this rigorous benchmark study.
#QuantumMachineLearning #HighEnergyPhysics #Research
#QuantumMachineLearning #HighEnergyPhysics #Research
QFI-optimized quantum reservoirs outperform tuned classical networks by 25–69% on Mackey–Glass forecasting. A fundamental bound connects QFI to loss gradients through measurement, readout, and task factors, identifying optimization bottlenecks.
#QuantumMachineLearning #QuantumOptimization #Research
#QuantumMachineLearning #QuantumOptimization #Research
Quantum Fisher Information Optimization Enables Reservoir Advantage Over Classical Networks
arxiv.org
October 14, 2026 at 12:12 PM
QFI-optimized quantum reservoirs outperform tuned classical networks by 25–69% on Mackey–Glass forecasting. A fundamental bound connects QFI to loss gradients through measurement, readout, and task factors, identifying optimization bottlenecks.
#QuantumMachineLearning #QuantumOptimization #Research
#QuantumMachineLearning #QuantumOptimization #Research
Theoretical framework establishes rigorous generalization error bounds for Transformer-based neural quantum states under in-context learning, proving prediction error decreases inversely with context examples and network depth.
#QuantumMachineLearning #NeuralQuantumStates #TransformerTheory
#QuantumMachineLearning #NeuralQuantumStates #TransformerTheory
Transformer-Based Neural Quantum States: Generalization Bounds via In-Context Learning
arxiv.org
October 13, 2026 at 3:01 PM
Theoretical framework establishes rigorous generalization error bounds for Transformer-based neural quantum states under in-context learning, proving prediction error decreases inversely with context examples and network depth.
#QuantumMachineLearning #NeuralQuantumStates #TransformerTheory
#QuantumMachineLearning #NeuralQuantumStates #TransformerTheory
Purdue University is hiring:
Postdoctoral Research Associate in Quantum Machine Learning
#postdoc #computerscience #quantummachinelearning
Postdoctoral Research Associate in Quantum Machine Learning
#postdoc #computerscience #quantummachinelearning
October 10, 2026 at 10:23 AM
Purdue University is hiring:
Postdoctoral Research Associate in Quantum Machine Learning
#postdoc #computerscience #quantummachinelearning
Postdoctoral Research Associate in Quantum Machine Learning
#postdoc #computerscience #quantummachinelearning