"Design and implementation of machine learning based tools and workflows for biomedical data analysis"
Abstract
There are many challenges in biomedical sciences, where complexity, heterogeneity, and other characteristics of big data are rather common and are often accompanied by a lack of fundamental understanding of the underlying systems and processes. Machine learning technologies are increasingly used for data analysis in complex systems such as medical diagnostic systems. This work will mainly describe machine learning as a computational and analytical paradigm in the domain of biomedical data science. Specific tasks will include visualization and exploratory data analysis, clustering, classification with a focus on 3D microscopic image processing and quick-and-easy data exploration on the web. Two projects will be discussed in detail. The first uses machine learning for segmentation and morphometry-based classification with the ability to analyze gigabytes of imaging data in a high-throughput fashion. The second covers the design and development of a scalable web-based analytical toolbox to build and run in-browser applications for interactive data wrangling, visualization, and analysis including machine learning algorithms.
