Advanced Machine Learning Course

“Advanced Machine Learning” course (ECE-GY 7143) at NYU Tandon by Anna Choromanska

The “Advanced Machine Learning” course (ECE-GY 7143) at NYU Tandon by Anna Choromanska offers a deeper dive into the topics learned in the Machine Learning course from Fall 2023 semester.

Course Overview

It begins with foundational concepts such as the perceptron, exponentiated gradient, and online multi-class classification, setting the stage for understanding linear classifiers and gradient-based optimization. The course then delves into deep learning, covering feed-forward neural networks, back-propagation, convolutional neural networks (CNNs), regularization techniques, adversarial networks, auto-encoders, LISTA, transformers, and the complexities of non-convex optimization landscapes. Students will explore empirical risk minimization and PAC bounds to grasp performance guarantees in learning theory. The curriculum also includes topics like Occam’s Razor, VC dimension, support vector machines (SVM), and kernels, followed by a thorough introduction to optimization methods such as gradient descent, stochastic gradient descent, and more advanced techniques like BFGS, LBFGS, ISTA, and FISTA. The course covers expectation maximization (EM), the curse of dimensionality, the Johnson–Lindenstrauss lemma, principal component analysis (PCA), and various clustering algorithms including k-means and hierarchical clustering. The course concludes with student project presentations.

Review

The course stands out for its focus on cutting-edge research, with students learning directly from seminal research papers, which keeps the content current and relevant. Prof. Anna has a unique way of making complex theories accessible and engaging, often incorporating humor into her explanations, which greatly enhances the learning experience. Unlike courses centered on deep learning applications, this course delves deeply into the theoretical foundations of machine learning, providing a robust understanding that is essential for anyone looking to advance in this field. I highly recommend it to anyone who wants to build on their machine learning knowledge.