Thema der Dissertation:
Cell - type annotation in single-cell chromatin accessibility data Thema der Disputation:
Nonlinear Dimension Reduction Techniques: MDS, t-SNE, and UMAP
Cell - type annotation in single-cell chromatin accessibility data Thema der Disputation:
Nonlinear Dimension Reduction Techniques: MDS, t-SNE, and UMAP
Abstract: High-dimensional data present substantial challenges, especially in terms of the computational resources they require and the difficulty in understanding and visualizing their complex structures. As a result, reducing the dimension of the data while preserving its essential features is of great interest from various perspectives. While traditional linear methods like principal component analysis (PCA) are commonly applied to reduce dimension, they often fall short in capturing the nonlinear data patterns. This is where nonlinear dimension reduction techniques, also known as manifold learning, come into play, offering a way to map high-dimensional data into a more comprehensible, lower-dimensional form. This transformation facilitates a clearer data visualization and enhances interpretability. In this talk I will explore three popular nonlinear dimension reduction methods: multidimensional scaling (MDS), t-distributed stochastic neighbor embedding (t-SNE), and uniform manifold approximation and projection (UMAP), and provide an intuitive understanding of their mechanisms.
The second part of the presentation will focus on my thesis work, which addresses cell-type annotation in scATAC-seq data.
The second part of the presentation will focus on my thesis work, which addresses cell-type annotation in scATAC-seq data.
Zeit & Ort
11.02.2025 | 09:00
Seminarraum 006
(Fachbereich Mathematik und Informatik, Königin-Luise-Str.24-26, 14195 Berlin)