Ensemble learning, the closer look
Ensembles are everywhere in Machine Learning. The RandomForest algorithm is arguably one of the most popular and successful Machine Learning approaches of all time. Dropout – famous regularisation strategy can be thought of as a bagging ensemble of many Deep Neural Networks. XGBoost acquired a reputation of a silver bullet in Kaggle competitions. Despite their immense popularity, ensemble methods remain a mystery for many. Their seeming incomprehensive complexity intimidates ML practitioners, preventing them from utilizing ensembles at their full capacity. In this lecture, we will break down and carefully examine all the most important ensemble learning concepts, such as bagging, boosting, stacking, and blending. We will use small examples to demonstrate the inner workings of each ensemble type. Participants will be able to put their ensemble learning understanding into practice in a Colab notebook.
- Python via Collaboratory
- Knowledge of Python (numpy, pandas, scikit-learn)
- Understanding of the basic machine learning (around supervised learning and overfitting)
Level of complexity of course
Mr. Dmytro Fishman
Dmytro Fishman is an Assistant of Data Science, his research is focused on applying machine learning and data mining methods to biological data. For his Ph.D. thesis, Dmytro is building an automatic tool for analyzing protein microarray experiments in immunological studies. He and his colleagues use Deep Learning on various biological data, including genomic data and microscopy images. Also, Dmytro has experience teaching various machine learning related subjects both at the University of Tartu and as an invited lecturer in private companies. He has served as a lecturer in Machine Learning at Ukrainian Catholic University and Kyiv School of Economics.
Fields of interests: Bioinformatics, Machine Learning, Deep Learning, Artificial Intelligence
Contacts: [email protected]