Deep Learning and Applications II

1 credits · Graduate

Topics include spectral representation, long memory processes “shallow models”: ARMA, filter banks, SVR/SVM, random forests and Probabilistic graphical networks; deep models: RNNs and CNNs for sequential modeling, attention networks deep learning frameworks, basic models and causal loss functions for financial time-series prediction. Distributed representations of discrete entities and applications in Natural Language Processing data and model fusion strategies, irregular time series low cost modeling strategies (model compression, cascades and low rank modeling).