提出了多示例嵌入学习(multi-instance learning,MIL)的实例关联性挖掘与强化算法(multi-instance embedding learning with instance affinity mining and reinforcement,MEMR),包括3个技术。关联性挖掘技术基于自定义的关联性指标,首...提出了多示例嵌入学习(multi-instance learning,MIL)的实例关联性挖掘与强化算法(multi-instance embedding learning with instance affinity mining and reinforcement,MEMR),包括3个技术。关联性挖掘技术基于自定义的关联性指标,首先在负实例空间中选择初始负代表实例集,然后根据正、负实例间的差异性,选择初始正代表实例集。关联性强化技术分别评估初始正、负代表实例集与整个实例空间的正负关联性,获得整体关联性更强的代表实例集。包嵌入技术通过嵌入函数将包转换为单向量进行学习。实验在4类应用领域和7种对比算法上进行。结果表明,MEMR的准确性总体优于其他对比算法,特别是在图像检索和网页推荐数据集上具有显著优势。展开更多
We study a simplified version of the Sachdev-Ye-Kitaev(SYK) model with real interactions by exact diagonalization. Instead of satisfying a continuous Gaussian distribution, the interaction strengths are assumed to be ...We study a simplified version of the Sachdev-Ye-Kitaev(SYK) model with real interactions by exact diagonalization. Instead of satisfying a continuous Gaussian distribution, the interaction strengths are assumed to be chosen from discrete values with a finite separation. A quantum phase transition from a chaotic state to an integrable state is observed by increasing the discrete separation. Below the critical value, the discrete model can well reproduce various physical quantities of the original SYK model,including the volume law of the ground-state entanglement, level distribution, thermodynamic entropy,and out-of-time-order correlation(OTOC) functions. For systems of size up to N=20, we find that the transition point increases with system size, indicating that a relatively weak randomness of interaction can stabilize the chaotic phase. Our findings significantly relax the stringent conditions for the realization of SYK model, and can reduce the complexity of various experimental proposals down to realistic ranges.展开更多
文摘提出了多示例嵌入学习(multi-instance learning,MIL)的实例关联性挖掘与强化算法(multi-instance embedding learning with instance affinity mining and reinforcement,MEMR),包括3个技术。关联性挖掘技术基于自定义的关联性指标,首先在负实例空间中选择初始负代表实例集,然后根据正、负实例间的差异性,选择初始正代表实例集。关联性强化技术分别评估初始正、负代表实例集与整个实例空间的正负关联性,获得整体关联性更强的代表实例集。包嵌入技术通过嵌入函数将包转换为单向量进行学习。实验在4类应用领域和7种对比算法上进行。结果表明,MEMR的准确性总体优于其他对比算法,特别是在图像检索和网页推荐数据集上具有显著优势。
基金Project supported by the National Basic Research Program(973 Program) of China(No.2011CB706506)the National Natural Science Foundation of China(Nos.51221004 and 51375012)+1 种基金the National High-Tech R&D Program(863 Program) of China(Nos.2013AA041303 and 2013IM030500)the Zhejiang Provincial Natural Science Foundation of China(No.Y13E050014)
基金This work was supported by the National Natural Science Foundation of China(11434011,11522436,11774425,11704029)the National Key R&D Program of China(2018YFA0306501)+1 种基金the Beijing Natural Science Foundation(Z180013)the Research Funds of Renmin University of China(16XNLQ03 and 18XNLQ15)。
文摘We study a simplified version of the Sachdev-Ye-Kitaev(SYK) model with real interactions by exact diagonalization. Instead of satisfying a continuous Gaussian distribution, the interaction strengths are assumed to be chosen from discrete values with a finite separation. A quantum phase transition from a chaotic state to an integrable state is observed by increasing the discrete separation. Below the critical value, the discrete model can well reproduce various physical quantities of the original SYK model,including the volume law of the ground-state entanglement, level distribution, thermodynamic entropy,and out-of-time-order correlation(OTOC) functions. For systems of size up to N=20, we find that the transition point increases with system size, indicating that a relatively weak randomness of interaction can stabilize the chaotic phase. Our findings significantly relax the stringent conditions for the realization of SYK model, and can reduce the complexity of various experimental proposals down to realistic ranges.
