What's the Connection? Jan 1, 2017 Link to preprint (bioRxiv) An unsupervised study of problem solving In this problem-solving study, we used unsupervised learning to explore the rapid interaction between competing brain networks during complex task performance. Methods: Intracranial EEG (electrophysiological recordings) Signal processing Time-series analysis Frequency domain analysis Dimensionality reduction (PCA, t-SNE, feature engineering) Clustering Scripting in MATLAB & EEGLAB Meir Meshulam Postdoc researcher Meir Meshulam is a postdoc in Neuroscience at Princeton University, using machine learning and deep learning to study human cognition.