University of Kentucky

MA 721: Topics in Numerical Analysis: Deep Learning

Fall 2023

List of Reading for Class Presentations

Please prepare slides for your presentation and submit it by email a week before the planned presentation date for comments/grading.

Paper Presenter Week
Different activation function and optimizer:
1. Maxout Networks
2. ADADELTA: An Adaptive Learning Rate Method
Initializers:
1. Understanding the difficulty of training deep feedforward neural networks
2. Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification		 Ali Alsetri 12/1
Generalization errors of optimizer:
1. The Marginal Value of Adaptive Gradient Methods in Machine Learning
2. When do adaptive optimizers fail to generalize?
Different local minimum and generalization:
1. On Large-Batch Training for Deep Learning: Generalization Gap and Sharp Minima
2. Sharp Minima Can Generalize For Deep Nets		 Kotaro Kajita 11/20
Adam optimizer:
Adam: A Method for Stochastic Optimization		 Alex Emmons 11/29
CNN Visualization:
Visualizing and Understanding Convolutional Networks		 Joshua Peterson 11/27