DS 303 - Course Syllabus

(TENTATIVE)

Module	Weeks	Topics and Keywords
	1	citizens of ML and the key challenge of ML
I		learning as constraint satisfaction
	2	geometry of pseudo-inverse, overparameterization
	3	perceptron, max-margin classification, support vector machines (SVM)
II		learning in the presence of noise, generalization, & regularization
	4	ridge regression, Quiz 1
	5	bias-variance tradeoff
	6	geometry of soft-margin SVM
	7	kernel methods & reproducing kernel hilbert spaces
	8	Mid-semester exam
III	9	clustering & dimensionality reduction
IV	10	probabilistic modeling: discriminative v/s generative, LDA, QDA
V	11	loss-functions, risk, empirical risk minimization
	12	generalization gap Quiz 2
VI		advanced topics
	13	unsupervised learning & generative modelling
	14	optimization, stochastic gradient descent, non-convexity
	15	robustness, and non-euclidean spaces
	16	End-semester exam