Dr. Kosmas Kepesidis

Data analytics

Exploring the application of molecular spectroscopy to cancer diagnosis using machine learning. Working on developing a new, non-invasive and rapid way of diagnosing cancer. Exploring signal pre-processing methods to remove noise and identify signals of interest. Building and optimizing classifiers using data mining and machine learning methods. Identifying signal features for achieving clinically meaningful predictions.

Further development of a Web Crawler with Python for collecting data from comparison websites. Image binarization with classical methods (Otsu’s algorithm, adaptive thresholding) and Deep Learning (CNN). Text classification with machine-learning methods (logistic regression, SVM and LightGBM) and Deep Learning methods (Word Embedding, RNN).

Work within a data-science team, performing text mining and document classification. For this project, various classical machine-learning and text-mining methods were used, such as text pre-processing, bag of words model, logistic regression, SVM, cross validation etc.

Research - Theoretical Quantum Physics - Nano-Mechanical Systems

One-semester training with an estimated workload of 200 hours covering theoretical and practical concepts in the field of Data Science and its potential in business use cases.

Theoretical Quantum Optics

Theoretical Quantum Optics