Through incremental integration and independent research and development, build a method library of big data quality control, automatic modeling and analysis, data mining and interactive visualization, form a tool library with high reliability, high scalability, high efficiency and high fault tolerance, realize the integration and sharing of collaborative analysis methods of multi-source heterogeneous, multi-granularity, multi-phase, long-time series big data in three pole environment, as well as high Efficient and online big data analysis and processing.
NMF is a matrix factorization method proposed by Lee and Seung in Nature in 1999. It makes all decomposed components nonnegative and realizes nonlinear dimension reduction at the same time. NMF has gradually become as one of the most popular multidimensional data processing tools in the fields of signal processing, biomedical engineering, pattern recognition, computer vision and image engineering.
ICA is a technique for separating source signals from linear mixtures of multiple source signals. There is no prior knowledge except that the source signal is known to be statistically independent. Compared with the traditional filtering method and the cumulative average method, ICA can eliminate the noise while hardly destroying the details of other signals, and the denoising performance is often much better than the traditional filtering method. Moreover, compared with traditional signal separation methods based on feature analysis, such as singular value decomposition (SVD) and principal component analysis (PCA), ICA is an analysis method based on high-order statistical characteristics. In many applications, the analysis of higher-order statistical properties is more practical.