Upcoming Seminars

Speaker

Dr. Ross Mawhorter, Postdoctoral Scholar at the UofA, supervised by Dr. Matthew Guzdial

Title

The Very Hungry Researcher: Form and Function in Procgen Research

Abstract
Procedural Content Generation, or procgen, is a research field that studies methods for automatically producing all kinds of different designs. In my research, I focus on procedural level design for videogames. While there are many different methods for procedural level design, spanning essentially the whole field of AI (from constraint solving to reinforcement learning to machine learning), all of these methods leave something to be desired. There is a tendency to conflate form (the actual content being generated) and function (the player experiences that the content supports). In this talk, I will survey the field of procgen, and describe how the existing methods handle this dichotomy. I will explain where progress has been made, but I will also describe the ways that these contributions still leave me hungry.

Presenter Bio
Ross Mawhorter is a recent PhD graduate from the University of California, Santa Cruz. His passion for research led him to abandon sunny California for the frozen north as a postdoc in Matthew Guzdial's GRAIL lab. He studies procgen, working in his free time on generating Super Metroid ROMhacks. Along with procedural level design, his academic interests include complexity theory and cophylogeny.

Timing & Location

UComm Seminar Room 2-108

pizza from 11:30, seminar from noon to 1

Add event to calendar

Add event

Speaker

Rohan Saha, PhD student, University of Alberta, supervised by Dr. Alona Fyshe

Title

Understanding the Staged Dynamics of Transformers in Learning Latent Structure

Abstract
Language modeling has shown us that transformers can discover latent structure from context, but the dynamics of how they acquire different components of that structure remain poorly understood, leading to assertions that models just remix training data. In this work, we use the Alchemy benchmark in a controlled setting (Wang et al., 2021) to investigate latent structure learning. We train a small decoder-only transformer on three task variants: 1) inferring missing transitions from partial contextual information, 2) composing simple rules to solve multi-transition sequences, and 3) decomposing complex multi-step examples to infer intermediate transitions. By factorizing each task into interpretable components, we show that the model learns the different latent structure components in discrete stages. We also observe an asymmetry: the model composes fundamental transitions robustly, but struggles to decompose complex examples to discover the atomic transitions. Finally, using causal interventions, we identify layer-specific plasticity windows during which freezing substantially delays or prevents stage completion. These findings provide insight into how a transformer model acquires latent structure, offering a detailed view of how capabilities evolve during training.

Presenter Bio
Rohan Saha is a PhD student at the University of Alberta studying the learning dynamics of transformer models. 

Timing & Location

UComm Seminar Room 2-108

pizza from 11:30, seminar from noon to 1

Add event to calendar

Add event

NO SEMINAR - Good Friday

Speaker

Dr. Bahareh Tolooshams, Amii Fellow and Assistant Professor in the Electrical & Computer Engineering Department at the University of Alberta

Title

TBA

Abstract
TBA

Presenter Bio
TBA

Timing & Location

Amii HQ 2nd floor event space (10065 Jasper Ave)

pizza from 11:30, seminar from noon to 1

Add event to calendar

TBA

Speaker

Calarina Muslimani, PhD student at the University of Alberta, supervised by Dr. Matthew Taylor

Title

Reward Design and Evaluation in Reinforcement Learning

Abstract
This talk will focus on the design and evaluation of reward functions in reinforcement learning. I will begin by discussing the key challenges in reward design and the difficulties in determining whether a reward function is properly specified. Then, I’ll introduce an approach to support RL practitioners in designing more effective and aligned reward functions.

Presenter Bio
Calarina (Callie) Muslimani is a fourth-year PhD student at the University of Alberta in the Reinforcement Learning and Artificial Intelligence (RLAI) Lab, advised by Matthew E. Taylor. Her research focuses on designing human-aligned reward functions for reinforcement learning, including developing metrics to evaluate reward functions and creating reward learning algorithms. 

Timing & Location

UComm Seminar Room 2-108

pizza from 11:30, seminar from noon to 1

Add event to calendar

Add event

Speaker

TBA

Speaker

Dr. Amber Simpson, Amii Fellow and Professor, Faculty of Medicine & Dentistry - Radiology & Diagnostic Imaging Department

Title

TBA

Abstract
TBA

Presenter Bio
TBA

Website

TBA

Timing & Location

Amii HQ 2nd floor event space (10065 Jasper Ave)

pizza from 11:30, seminar from noon to 1

Add event to calendar

TBA

Speaker

Dr. Liam McCoy, Neurology Resident Physician, University of Alberta, and Research Affiliate, Massachusetts Institute of Technology, hosted by Dr. Randy Goebel

Title

TBA

Abstract
TBA

Presenter Bio
TBA

Timing & Location

UComm Seminar Room 2-108

pizza from 11:30, seminar from noon to 1

Add event to calendar

TBA

Speaker

Jiamin He, PhD student at the University of Alberta, supervised by Dr. Martha White

Title

TBA

Abstract
TBA

Presenter Bio
TBA

Timing & Location

UComm Seminar Room 2-108

pizza from 11:30, seminar from noon to 1

Add event to calendar

TBA

NO SEMINAR - Upper Bound

Speaker

TBA