Spring 2025 talk schedule
Date Talk by February 14 Christine Lee February 28 Christine Lee March 28 Geonhee Cho April 11 Jeremiah Birrell April 25 Jeremiah Birrell
Author: vne11
Talk by Ivan Ojeda-Ruiz on Friday 4/5
Title: An Overview of Graph Neural Network Propagation Modules Abstract: Graph Neural Networks (GNNs) offer powerful methods to analyze the complex relationships inherent in graph-structured data. Propagation modules sit at the core of GNNs, driving the process of learning representations by aggregating information across nodes. This talk explores the landscape of GNN propagation modules. We’ll…
Talk February 9 by Suho Oh: Intro to graph neural networks
Abstract: We’ll go over graph neural networks (GNN) that has been gaining popularity over the last few years. We will cover the definition, some different models of the similar flavor, and if time permits some coding examples on how one can use them for your purpose. Resources: https://web.stanford.edu/class/cs224w/slides/03-GNN1.pdf here is the slides going over basics of…
Talk 12/1/2023 Impossibility of Fairness by Hiro Lee Tanaka and Jake Fillman
Abstract: This talk will not build on the past sessions of the seminar. We will prove that there is no way to assign risk scores or develop decision-making algorithms that satisfy basic axioms fairness. The talk will begin by giving some examples of (un)fairness in the real world. 2023.12.01 Fairness
11/6/2023 Difference between SGD and ADAM by Xiaoxi Shen and Jialong Li
Abstract: With the growth of dimensionality of the data we encounter nowadays, the classical gradient descent methods need to speed up. In the presentation, we will start by reviewing the gradient descent and stochastic gradient descent and discuss their theoretical guarantees. After that, more commonly used optimization algorithms in deep learning will be introduced and…
Talk 11/3/23 Gradient descent by Usufu Nyakoojo and Humu Mohammed
Abstract: In training an artificial neural network, we need to find model parameters that optimize the cost function while ensuring sufficient learning rate of the machine. Gradient descent and in particular the stochastic gradient descent, comes in handy as an iterative optimization approach to address this need. In this talk, we shall present how this can be achieved i.e. how the cost function…
Talk 10/6/2023 Artificial neural networks and universal approximation by Douglas Limmer
Abstract: Artificial neural networks are designed to mimic the way neurons in the brain interact with each other in an attempt to achieve a level of thinking similar to what human beings can attain. This talk looks at the beginnings of implementation of neural networks, the limitations and challenges neural networks have faced, and adjustments…