This Week in Mathematics


<< Back to Current Week [2024-09-15 - 2024-09-21]	>> Next Week [2024-09-22 - 2024-09-28]

<< View Previously Scheduled Events

Beyond Sep 28, 2024

Entries yielded: 0

Thursday October 03, 2024

Geometry and Topology Seminar [url]
TBA
- Kurt Vinhage, University of Utah
Time: 3:05 Room: TBA
Abstract/Desc: TBA

Thursday October 03, 2024

Geometry and Topology Seminar [url]
TBA
- Kurt Vinhage, University of Utah
Time: 3:05 Room: TBA
Abstract/Desc: TBA

Thursday October 03, 2024

Financial Mathematics Seminar [url]
TBA
- Ali Kara, Florida State University

Tuesday October 08, 2024

Geometry and Topology Seminar [url]
TBA
- Ali Alp Uzman, University of Utah
Time: 3:05 Room: 301
Abstract/Desc: TBA

Thursday October 10, 2024

Financial Mathematics Seminar [url]
TBA
- Ali Kara, Florida State University
Time: 3:05 Room: LOV231

Friday October 11, 2024

Mathematics Colloquium [url]
Exploiting Low-Dimensional Data Structures and Estimating Scaling Laws for Transformer Neural Networks
- Wenjing Liao, Georgia Tech
Time: 3:05 Room: Lov 101
Abstract/Desc: When training deep neural networks, a model’s generalization error is often observed to follow a power scaling law dependent on the model size and the data size. Perhaps the best-known example of such scaling laws is for transformer-based large language models (LLMs), where networks with billions of parameters are trained on trillions of tokens of text. One theoretical interest in LLMs is to understand why transformer scaling laws exist. To answer this question, we exploit low-dimensional structures in language datasets by estimating its intrinsic dimension, and establish statistical estimation and mathematical approximation theories for transformers to predict the scaling laws. By leveraging low-dimensional data structures, we can explain transformer scaling laws in a way which respects the data geometry. Furthermore, we test our theory with empirical observations by training LLMs on natural language datasets, and find strong agreement between the observed empirical scaling laws and our theoretical predictions.

Thursday October 17, 2024

Financial Mathematics Seminar [url]
TBA
- Munavar Ali, Florida State University
Time: 3:05 Room: LOV231

Wednesday October 30, 2024

Biomathematics Seminar
Explosive Opinion Polarization and Depolarization with Nonlinear Incidence
- Haoyang Qian, FSU
Time: 3:05 Room: 232 Love

Thursday October 31, 2024

Financial Mathematics Seminar [url]
TBA
- Ololade Sowunmi, Florida State University
Time: 3:05 Room: LOV231

Thursday November 07, 2024

Financial Math Seminar

- David Herzog, Iowa State University
Time: 3:05pm Room: Lov 231

Wednesday November 13, 2024

Biomathematics Seminar
How Reasoning About Errors can Affect Indirect Reciprocity
- Bryce Morsky, FSU
Time: 3:05 Room: 232 Love
Abstract/Desc: Indirect reciprocity serves as a pivotal mechanism in fostering cooperative behaviour within social groups. However, challenges arise in maintaining its effectiveness due to errors in assessments and actions, leading to disagreements on reputations and an unravelling of cooperation. To address these challenges, we explore two cognitive mechanisms, Bayesian and abductive reasoning. By considering the probabilities of errors along with prior beliefs about the prevalence of good and bad individuals, observers employ reasoning to aide in evaluating the reputations of others. Under Bayesian reasoning, this is an application of Bayes' rule. While under abductive reasoning, individuals' beliefs and observations are combined to infer reputations. Further, abductive reasoning is in a sense "lazy" since it only considers the simplest explanations when accounting for errors. We assess the effectiveness of such reasoning on promoting cooperation under various conditions. We find that Bayesian reasoning generallyundermines cooperation relative to the non-reasoning and abductive reasoning models, expect for the Scoring norm. Abductive reasoning generally outperforms the non-reasoning and Bayesian reasoning models but may undermine cooperation in high-error situations. Further, we find that a slight pessimism about the reputations of others or a resistance to changing reputations can benefit cooperation.

Thursday November 14, 2024

Financial Mathematics Seminar [url]
TBA
- Ryan Bausback, Florida State University
Time: 3:05 Room: LOV231

Thursday November 21, 2024

Financial Mathematics Seminar [url]

- Gu Wang, WPI
Time: 3:05-3:55 Room: LOV 231

Thursday December 05, 2024

Financial Mathematics Seminar [url]
TBA
- Farez Siddiqui, Florida State University
Time: 3:05 Room: LOV231

Problems? Email webmaster@math.fsu.edu.

Department of Mathematics

This Week in Mathematics