Empirical Quantum Advantage Analysis of Quantum Kernels for Gene Expression Classification: A Comparative Study with Classical Approaches
This research paper investigates the potential of quantum computing for gene expression classification, comparing the performance of quantum and classical kernels in feature selection and classification accuracy, ultimately suggesting that quantum kernels may offer advantages in specific configurations.
양자 오토인코더 기반 테스트 시간 학습: 데이터 분포 변화 및 노이즈가 있는 양자 회로에 대한 강건성 향상
본 논문에서는 테스트 시간 학습(Test-Time Training, TTT)을 양자 오토인코더(Quantum Auto-Encoder, QAE)와 결합한 QTTT(Quantum Test-Time Training) 프레임워크를 제안하여, 훈련 데이터와 테스트 데이터 간의 분포 변화 및 양자 회로 오류 문제를 해결하고자 합니다.
Test-Time Training with Quantum Auto-Encoders for Improved Quantum Neural Network Generalization and Noise Mitigation
Test-time training with quantum auto-encoders (QTTT) is a novel approach to improve the generalization and noise resilience of quantum neural networks (QNNs) by fine-tuning model parameters during inference using a self-supervised learning objective.
The Impact of Limited Measurement Shots on Entanglement Classification in Quantum Machine Learning
Even for simple quantum problems with known solutions, like classifying entanglement, the limitations of quantum measurements can significantly hinder the performance of quantum machine learning algorithms, particularly when the number of measurement shots is limited.
Trainability and Expressivity of Quantum Re-uploading Models: An Analysis of Gradient Magnitudes and Frequency Profiles
Quantum re-uploading (QRU) models, while potentially powerful for machine learning, exhibit inherent limitations in their ability to be trained effectively and represent complex functions due to the behavior of their gradients and the vanishing of high-frequency components in their output.
When Quantum Models Outperform Classical Counterparts: Exploring Learning Beyond Minimum Norm Least Squares
Variational Quantum Circuits (VQCs) have the potential to surpass classical machine learning models by achieving weight vector norms unattainable by classical algorithms like Minimum Norm Least Squares (MNLS), particularly in high-dimensional feature spaces.
Implementation and Evaluation of a Quantum Projective Simulation Agent on a Photonic Quantum Computer for a Transfer Learning Task
This research demonstrates the first physical implementation of a quantum projective simulation (PS) agent on a photonic quantum computer, showcasing its ability to outperform classical PS agents in a specific transfer learning task by leveraging quantum effects.
Quantum Algorithm for Sparse Online Learning with Truncated Gradient Descent: Achieving Quadratic Speedup in Dimension While Maintaining Regret Bounds
This paper presents a quantum algorithm for sparse online learning that achieves a quadratic speedup in the dimension of the data compared to classical counterparts, while maintaining a similar regret bound, making it particularly suitable for high-dimensional learning tasks.
ShadowGPT: Using Machine Learning to Predict Ground State Properties of Quantum Many-Body Systems from Randomized Measurement Data
ShadowGPT, a generative pre-trained transformer model, can accurately predict ground state properties of quantum many-body systems by learning from randomized measurement data, offering a potential solution to complex quantum problems using classical machine learning.
Quantum-Enhanced Multiclass Support Vector Machines with Quantum Advantage for Real-World Applications
Quantum machine learning algorithms, specifically those leveraging quantum kernels, can outperform classical methods in multiclass classification tasks, demonstrating a potential quantum advantage.
© 2024 by Linnk AI