This respository contains:
- Personal Summary of the whole course [PDF]
- Code for the projects (Counted 30% to our final grade)
All of these tasks were related to health care and biology.
Final Project Grade: 5.63
Outlieder Detection. The features in the given dataset represent volumes of surface areas of various regions of the brain. We then try to solve a regression task, however the dataset has outliers and bad features.
Our Approach: Use a Random Forest to weight feature importance
Class Imbalance. We get a Dataset that is heavely imbalanced and need to regress values (supervised task).
Our Approach: SVM with class-weights
ECG anomaly detection. As input we get a timeseries of ECG values. OUr task it to classify if the patient of this ECG has an anomaly (Decide between 3 possible anomalies) or has a normal heartbeat.
Our Approach: Feature Extraction with libraries and personal Algorithms to extract Q,R,S point etc. + Gradient Boosting
Sleep detection for mice. As input we get EEG and EMG signals of a mouse for 24 hours. We then need to decide when the mouse slept and when it was awake.
Our Approach: Fourier Transformation weith Hamming windows + CNN with Softmax + Postprocessing with a Dense Neural Net