Traditionally, in supervised machine learning, (a significant) part of the available data (usually 50%-80%) is used for training and the rest—for validation. In many problems, however, the data are highly imbalanced in regard to different classes or does not have good coverage of the feasible data space which, in turn, creates problems in validation and usage phase. In this paper, we propose a technique for synthesizing feasible and likely data to help balance the classes as well as to boost the performance in terms of confusion matrix as well as overall. The idea, in a nutshell, is to synthesize data samples in close vicinity to the actual data samples specifically for the less represented (minority) classes. This has also implications to the so-called fairness of machine learning. In this paper, we propose a specific method for synthesizing data in a way to balance the classes and boost the performance, especially of the minority classes. It is generic and can be applied to different base algorithms, for example, support vector machines, k-nearest neighbour classifiers deep neural, rule-based classifiers, decision trees, and so forth. The results demonstrated that (a) a significantly more balanced (and fair) classification results can be achieved and (b) that the overall performance as well as the performance per class measured by confusion matrix can be boosted. In addition, this approach can be very valuable for the cases when the number of actual available labelled data is small which itself is one of the problems of the contemporary machine learning. 相似文献
Russian Journal of Non-Ferrous Metals - Rolling bearing rings comprise a large reserve for expanding powder metallurgy production. This reserve is being incompletely implemented. The hot forging of... 相似文献
Journal of Communications Technology and Electronics - Abstract—The two-dimensional diffraction problem of a plane polarized TM-wave on two identical solid silver nanocylinders is considered.... 相似文献
The efficiency of any electric cell or battery is very important. To keep it in mind it has been studied the columbic efficiency, voltaic efficiency and energy efficiency of a PKL (Pathor Kuchi Leaf) Quasi Voltaic Cell or Modified Voltaic Cell. It was found that the columbic efficiency data illustrated that this efficiency was lower comparing to other efficiencies may be the absence of salt bridge or separator between the electrodes. Because, our designed and fabricated PKL cell does not have any salt bridge. So that the internal resistance is lower than the traditional voltaic cell and as a result more current was found. The voltage and current changes with time and I–V characteristics for PKL unit cell, module, panel and array have also been studied. It is shown that the voltaic and energy efficiency have been studied. However, the highest efficiency was obtained for 40% PKL sap with 5% secondary salt in 55% aqueous solution, which implies that the concentration of PKL juice can play an important role regarding efficiency. It was also found that the average energy efficiency was 97.43% and it was also found that the average voltaic efficiency was 57.29%. Finally, morphological study FESEM (Field Emission Scanning Electron Microscopy) has also been performed. It is seen that the results confirmed that Zn was deposited on the Cu surface during the electro deposition process in PKL solution. Using AAS, it has been measured the concentration of [Cu2+] as a reactant ion and the concentration of [Zn2+] as a product ion those have been tabulated and graphically discussed. The variation of pH has also been studied with time and which was also tabulated and graphically discussed.
Photonic Network Communications - Optical packet switching is a prominent technology proposing not only a reduction in the energy consumption by the elimination of numerous... 相似文献
RCAS1 is a protein that participates in regulation of the tumor microenvironment and its immune responses, all in order to evade the immune system. The aim of this study was to analyze RCAS1 expression in urothelial bladder cancer cells (and in fibroblasts and macrophages of the tumor stroma) and its relationship with the histological pattern of malignancy. Eighty-three postcystectomy patients were enrolled. We analyzed the histological maturity (grade), progress (pT stage), tissue invasion type (TIT), nonclassic differentiation number (NDN), and the ability to metastasize (pN). The expression of RCAS1 protein was analyzed by immunohistochemistry. Indicators of histological malignancy were observed solely in association with the RCAS1 expression in cells in the border parts (BPs) of the tumor. Histological malignancy of the tumor, indicated by the pT and pN, and metastasis-free survival time, correlated significantly with RCAS1 expression in tumor neoplastic cells, whereas malignancy determined by grade, TIT, and NDN correlated with RCAS1 expression in fibroblasts and macrophages in the tumor microenvironment. These findings suggest that the increased RCAS1 expression depends on its cellular source and that RCAS1 expression itself is a component of various signaling pathways. The immune escape occurs within the tumor BPs, where the increase in the RCAS1 expression occurs within tumor cells and stromal cells in its microenvironment. We conclude that the histological pattern of tumor malignancy, indicated by grade, TIT, NDN, pT, and pN is a morphological indicator of immune escape. 相似文献
Stability constants and formation enthalpies of supramolecular complexes of crown ethers and their cyclic and acyclic analogues are determined on the basis of experimental data obtained by different physicochemical methods in the terms of a general approach developed and implemented in the ChemEqui software package. The established regularities of variation of stability of complexes are discussed as dependent on the ligand structure, nature of the cation, solvent, and anion. The applicability of the suggested method of determining complexation selectivity is shown for multicomponent equilibria in solutions.
