Machine learning in Hilbert spaces has become a useful modeling and prediction tool in many areas of science and engineering, says Xu and Ye, and there has also been an emerging interest in developing learning algorithms in Banach spaces. They contend that just as machine learning is usually well-posed in reproducing kernel Hilbert, so it is desirable to solve learning problems in Banach spaces endowed with certain reproducing kernels. Though a concept of reproducing kernel Banach spaces in the context of machine learning by employing the notion of semi-inner productions, has appeared, they find that this use of semi-inner product has its limitations. Therefore, in this paper they systematically study the construction of reproducing kernel Banach spaces without using semi-inner products. Annotation ©2019 Ringgold, Inc., Portland, OR (protoview.com)
