Thema der Dissertation:
Dynamical Aspects of the Evolution of Segmental Duplications in the Human Genome Thema der Disputation:
The Support Vector Machines (SVMs) and the kernel method
Dynamical Aspects of the Evolution of Segmental Duplications in the Human Genome Thema der Disputation:
The Support Vector Machines (SVMs) and the kernel method
Abstract: The support vector machine (SVM) is a widespread method applied for classification tasks. There are several characteristics of the method which deserve a special interest. One of them is a good interpretability of the SVM, especially, when we consider it geometrically. Also some generalizations can be added to the method in order to make it applicable to a broader set of tasks, including multiclass classification problems, regression, and the identification of non-linear decision boundaries. In case of SVM, the use of the so-called "kernel trick" made support vector machines applicable for various classification tasks where classes are non-linearly separable in the original feature space. Finally, to understand the SVM one has to study concepts from linear algebra and optimization theory which have broad applications in other fields of mathematics.
I will focus on the mathematics behind the method, it's interpretation and parameters optimization. I will also discuss ways how SVM can be applied to multiclass classification tasks, how kernelization can be implemented into SVMs and advantages it brings.
I will focus on the mathematics behind the method, it's interpretation and parameters optimization. I will also discuss ways how SVM can be applied to multiclass classification tasks, how kernelization can be implemented into SVMs and advantages it brings.
Zeit & Ort
05.07.2022 | 15:30