Ilmira Terpugova

I worked on this multiclass classification problem for text articles with mutation description (Kaggle). I implemented the term-weighting scheme and centroid-based classifier which won the competition.

This project centered around the prediction of type II diabetes using genomic and epidemiological data collected from European residents. The project was challenging due to the large datasets where the number of features was much higher than the number of samples.

I implemented the distributed pipeline for the classification task to predict complex disease; e.g., type II diabetes which is believed to depend on the combination of several genes and lifestyle and environmental factors

This tool helps to identify individual right whales (Eubalaena glacialis) from aerial photographs using CNN architecture.

I implemented a type of protein classification using the word2vec model to transform protein primary structures. The project included two main parts: an unsupervised transformation of protein sequences of arbitrary lengths into fixed size real-valued vector and classification of the proteins using the transformed vectors.