摘要 :
As one of the important research topics in machine learning, loss function plays an important role in the construction of machine learning algorithms and the improvement of their performance, which has been concerned and explored ...
展开
As one of the important research topics in machine learning, loss function plays an important role in the construction of machine learning algorithms and the improvement of their performance, which has been concerned and explored by many researchers. But it still has a big gap to summarize, analyze and compare the classical loss functions. Therefore, this paper summarizes and analyzes 31 classical loss functions in machine learning. Specifically, we describe the loss functions from the aspects of traditional machine learning and deep learning respectively. The former is divided into classification problem, regression problem and unsupervised learning according to the task type. The latter is subdivided according to the application scenario, and here we mainly select object detection and face recognition to introduces their loss functions. In each task or application, in addition to analyzing each loss function from formula, meaning, image and algorithm, the loss functions under the same task or application are also summarized and compared to deepen the understanding and provide help for the selection and improvement of loss function.
收起
摘要 :
We argue that shocks to a housing market are transmitted through the hierarchy of quality tiers within a housing market. The result is the prediction of waves of house price changes accompanied by changes in transaction volume. Ou...
展开
We argue that shocks to a housing market are transmitted through the hierarchy of quality tiers within a housing market. The result is the prediction of waves of house price changes accompanied by changes in transaction volume. Our study is related to existing models of spatial ripple effects across housing markets. The data are from the Hong Kong housing market. The findings from Granger causality tests strongly support the argument that domino effects within a single housing market occur in response to external shocks.
收起
摘要 :
Two types of NiAlHf coatings doped with different content of Si (1 at.% and 2 at.%) were deposited on a Ni_3Al based single crystal superalloy IC32 by electron beam physical vapor deposition (EB-PVD) method, respectively. For comp...
展开
Two types of NiAlHf coatings doped with different content of Si (1 at.% and 2 at.%) were deposited on a Ni_3Al based single crystal superalloy IC32 by electron beam physical vapor deposition (EB-PVD) method, respectively. For comparison, NiAlHf coating with 0 at.% Si was also prepared. The oxidation tests were carried out at 1423 K in air. At the initial stage of oxidation, large amount of flake-like θ-Al_2O_3 was found on NiAlHf coating surface. However, no θ-Al_2O_3 was observed in 2 at.% Si doped NiAlHf coating except α-Al_2O_3. It revealed that the Si additions could contribute to the transformation from θ-Al_2O_3 to α-Al_2O_3 . When oxidation time prolonged to 100 h, it was found that the degradation of NiAlHf coating was very severe with no residual β-phase, which was due to the serious inter-diffusion between the coating and substrate. In contrast, the inter-diffusion in Si-doped coating was reduced with some residual p-phase and R-Ni(Mo, Re) precipitates. The presence of Si could retard the inter-diffusion of elements between coating and substrate, indicating a barrier diffusion effect. As a result, the oxidation resistance of NiAlHf coating was improved significantly.
收起
摘要 :
Electron energy loss spectroscopy (EELS) and energy dispersion spectroscopy (EDS) were used to study the chemical composition and bonding of grain boundary in CrMo steel. The bonding change in grain boundary of commercial CrMo ste...
展开
Electron energy loss spectroscopy (EELS) and energy dispersion spectroscopy (EDS) were used to study the chemical composition and bonding of grain boundary in CrMo steel. The bonding change in grain boundary of commercial CrMo steel is induced by P element segregation. The normalized 3d occupancies of states of irons both in bulk and at grain boundary were calculated according to the EELS spectra and they were related to the change in the chemical compositions of grain boundaries and the fracture mode. If the grain boundary has a higher occupancy of 3d state of iron than the bulk, the sample tends to have intergran-ular fracture, while if the grain boundary has almost the same occupancy of 3d state of iron as the bulk, the sample tends to have transgranular fracture.
收起
摘要 :
Page 15831. In Figure 3e, the key to the symbols, indicating different temperature conditions, was missing. The corrected figure is shown below: Page 15831. In Figure 4b, the X-axis was wrongly labeled as "Voltage (V)". The correc...
展开
Page 15831. In Figure 3e, the key to the symbols, indicating different temperature conditions, was missing. The corrected figure is shown below: Page 15831. In Figure 4b, the X-axis was wrongly labeled as "Voltage (V)". The correct label is "Wavelength (nm)". The corrected figure is shown below:
收起