janvargovsky/masters-thesis repository overview

⁠System for Support of Biomedical Data Analysis

The analysis of biomedical data is a current task, mainly thanks to the ever-evolving technologies for obtaining and preprocessing biological samples. The result is always real data suffering from many problems associated with mistakes, inaccuracies, incompleteness, etc.

This diploma thesis aims to prepare a software environment supporting experiments with real-world data using machine learning methods based on modern approaches (e.g. deep learning). This environment will be addressed to researchers and domain experts in biology and biochemistry specializing in the field of biomedical data analysis. The designed software environment will allow them to plan experiments and prepare data, configure and execute the experiment with results including a PDF report with tables and appropriate visualizations, save experiment configuration and results, and repeat the experiment.

Partial goals:

1. Review of systems, frameworks, and methods in the field of biomedical data analysis.
2. Design of system architecture, implementation of the user interface (the web is assumed), integration of the used frameworks (e.g. ML.NET, Accord.NET, Math.NET, ILNumerics, Tensorflow) and design of data storage.
3. Implementation and integration of selected methods, testing. Implementation of the methods with regards to thousands to tens of thousands of instances and tens to hundreds of attributes is expected. Experimental tasks in the area of classification, clustering, and feature selection are assumed.
4. Documentation of implemented methods and system components.