针对机器学习在流量异常检测中存在特征选择依赖经验、易受离群点影响导致鲁棒性差等问题,基于因素空间理论的“背景关系-背景分布-背景基”体系提出一种流量异常检测的基点分类方法。首先,数据预处理阶段使用KNN离群点检测算法去除数...针对机器学习在流量异常检测中存在特征选择依赖经验、易受离群点影响导致鲁棒性差等问题,基于因素空间理论的“背景关系-背景分布-背景基”体系提出一种流量异常检测的基点分类方法。首先,数据预处理阶段使用KNN离群点检测算法去除数据中的离群点,降低异常点对后续背景基提取的影响。其次,使用mRMR算法对数据特征进行排序,选择对分类最具影响力的特征标注为类别区分特征。然后,以内点判别法为理论基础优化背景基提取算法,提取训练数据中不同类别数据的背景基,得到各类别的单位认知包。最后,以单位认知包为核心构造基点分类算法(fundamental point classification algorithm,FPCA)实现异常流量的精准二分类。在NSL-KDD数据集上对所提方法的二分类实验准确率和F1-score分别达到92.48%和92.18%,检测性能优于同类型的其他机器学习方法。在CICIDS2017场景数据集上的测试进一步验证了所提方法在实际应用中的可行性。展开更多
We apply a Cross-Correlation （CC） method developed previously for detecting gamma-ray point sources to the WMAP first year data by using the Point-Spread Function of WMAP and obtain a full sky CC coefficient map. We...We apply a Cross-Correlation （CC） method developed previously for detecting gamma-ray point sources to the WMAP first year data by using the Point-Spread Function of WMAP and obtain a full sky CC coefficient map. We find that the CC method is a powerful tool to examine the WMAP foreground residuals which can be further cleaned accord- ingly. Evident foreground signals are found in the WMAP foreground cleaned maps and the Tegmark cleaned map. In this process 101 point sources are detected, and 26 of them are new sources additional to the originally listed WMAP 208 sources. We estimate the flux of these new sources and verify them by another method. As a result, a revised mask file based on the WMAP first year data is produced by including these new sources.展开更多
In quantum field theory, the physical vacuum, free from magnetic and electric fields (without regard to gravitational energy), is defined not as an empty space but as the ground state of the field consisting of quantu...In quantum field theory, the physical vacuum, free from magnetic and electric fields (without regard to gravitational energy), is defined not as an empty space but as the ground state of the field consisting of quantum harmonic oscillators (QHOs) characterized by zero-point energy. The aim of this work is to show that such physical vacuum may possess the properties similar to the properties of dark energy: the positive density, the negative pressure, and the possibility of so-called accelerated expansion. In the model discussed, the mass of QHOs determines the positive density of dark energy. The observed electric polarization of physical vacuum in an electric field means the existence of electric dipole moment of QHO, which, in turn, suggests the existence inside the QHO of a repulsive force between unlike charges compensating the attractive Coulomb force between the charges. The existence of such repulsive force may be treated as the existence of omniradial tensions inside every QHO. In terms of hydrodynamics, it means that the vacuum with this property may be regarded as a medium with negative pressure. The electric dipole-dipole interaction of QHOs under some condition may result in the expansion of physical vacuum consisting of QHOs. It is shown also that the physical vacuum consisting of QHOs is a luminiferous medium, and based on this concept the conditions are discussed for the emergence of invisiblity of any objects (in particular, dark matter). The existence of luminiferous medium does not contradict the second postulate of special relativity (the principle of constancy of the velocity of light in inertial systems), if to take into account the interaction of photons with QHOs and with virtual photons (the virtual particles pairs) created by quantum entities that constitute the inertial systems.展开更多
文摘针对机器学习在流量异常检测中存在特征选择依赖经验、易受离群点影响导致鲁棒性差等问题,基于因素空间理论的“背景关系-背景分布-背景基”体系提出一种流量异常检测的基点分类方法。首先,数据预处理阶段使用KNN离群点检测算法去除数据中的离群点,降低异常点对后续背景基提取的影响。其次,使用mRMR算法对数据特征进行排序,选择对分类最具影响力的特征标注为类别区分特征。然后,以内点判别法为理论基础优化背景基提取算法,提取训练数据中不同类别数据的背景基,得到各类别的单位认知包。最后,以单位认知包为核心构造基点分类算法(fundamental point classification algorithm,FPCA)实现异常流量的精准二分类。在NSL-KDD数据集上对所提方法的二分类实验准确率和F1-score分别达到92.48%和92.18%,检测性能优于同类型的其他机器学习方法。在CICIDS2017场景数据集上的测试进一步验证了所提方法在实际应用中的可行性。
基金Supported by the National Natural Science Foundation of China.
文摘We apply a Cross-Correlation （CC） method developed previously for detecting gamma-ray point sources to the WMAP first year data by using the Point-Spread Function of WMAP and obtain a full sky CC coefficient map. We find that the CC method is a powerful tool to examine the WMAP foreground residuals which can be further cleaned accord- ingly. Evident foreground signals are found in the WMAP foreground cleaned maps and the Tegmark cleaned map. In this process 101 point sources are detected, and 26 of them are new sources additional to the originally listed WMAP 208 sources. We estimate the flux of these new sources and verify them by another method. As a result, a revised mask file based on the WMAP first year data is produced by including these new sources.
文摘In quantum field theory, the physical vacuum, free from magnetic and electric fields (without regard to gravitational energy), is defined not as an empty space but as the ground state of the field consisting of quantum harmonic oscillators (QHOs) characterized by zero-point energy. The aim of this work is to show that such physical vacuum may possess the properties similar to the properties of dark energy: the positive density, the negative pressure, and the possibility of so-called accelerated expansion. In the model discussed, the mass of QHOs determines the positive density of dark energy. The observed electric polarization of physical vacuum in an electric field means the existence of electric dipole moment of QHO, which, in turn, suggests the existence inside the QHO of a repulsive force between unlike charges compensating the attractive Coulomb force between the charges. The existence of such repulsive force may be treated as the existence of omniradial tensions inside every QHO. In terms of hydrodynamics, it means that the vacuum with this property may be regarded as a medium with negative pressure. The electric dipole-dipole interaction of QHOs under some condition may result in the expansion of physical vacuum consisting of QHOs. It is shown also that the physical vacuum consisting of QHOs is a luminiferous medium, and based on this concept the conditions are discussed for the emergence of invisiblity of any objects (in particular, dark matter). The existence of luminiferous medium does not contradict the second postulate of special relativity (the principle of constancy of the velocity of light in inertial systems), if to take into account the interaction of photons with QHOs and with virtual photons (the virtual particles pairs) created by quantum entities that constitute the inertial systems.
