PROJECT TITLE :

Deep Cross-Output Knowledge Transfer Using Stacked-Structure Least-Squares Support Vector Machines

ABSTRACT:

This article introduces a new method for deep cross-output knowledge transfer that is based on least-squares support vector machines. The method is given the acronym DCOT-LS-SVMs. Its goal is to improve the generalizability of least-squares support vector machines (LS-SVMs) while avoiding the complicated parameter tuning process that occurs in many kernel machines. This will allow it to achieve its goal. The approach that has been suggested possesses two important characteristics: 1) DCOT-LS-SVMs was modeled after a stacked hierarchical architecture that combines several layer-by-layer LS-SVMs modules. This architecture served as an inspiration for DCOT-LS-SVMs. Both 1) cross-output knowledge transfer is used to leverage knowledge from the predictions of the previous module in order to improve the learning process in the current module, and 2) the module that forms the higher layer has additional input features that take into consideration all of the predictions made by the modules that came before it. By utilizing this strategy, the parameters of the model, which may include a tradeoff parameter C and a kernel width, may be arbitrarily assigned to each module in order to make the process of learning much more manageable. In addition to this, DCOT-LS-SVMs has the capacity to independently and swiftly decide the extent of the cross-output knowledge transfer between adjacent modules by utilizing a fast leave-one-out cross-validation strategy. In addition, we present an imbalanced version of DCOT-LS-SVMs, which we refer to as IDCOT-LS-SVMs. This is due to the fact that imbalanced datasets are frequently encountered in real-world scenarios. A comparison with five different methods of analysis on datasets from the University of California, Irvine, as well as a case study on the diagnosis of prostate cancer are used to illustrate how effective the proposed approaches are.