Machine Learning, Mathematics and Intuition

Abstract:
This talk introduces key concepts in Natural Language Processing (NLP) and computer vision with a focus on deep learning, starting from a comparison between simple linear regression models and more complex machine learning architectures. We will cover deep learning models such as Multi-Layer Perceptrons (MLPs), Convolutional Neural Networks (CNNs), and transformer architectures. The discussion will explore how words, images, and data points are represented as vectors, and how models learn to make predictions by minimizing loss functions through optimization techniques such as gradient descent and backpropagation. Essential mathematical foundations, including calculus, probability, statistics, and linear algebra, will be highlighted. Even if you haven't taken most of these courses, a basic understanding of derivatives as tools for minimization will help you grasp most of the talk. Additionally, this talk provides an overview of modern AI tools, pretrained models, and practical applications. It is designed as an accessible introduction for undergraduates interested in AI and its growing influence across various domains.