Upcoming Seminars

Speaker

Reginald McLean, Applied Research Scientist in Trust and Safety at Amii

Title

Towards a scientific understanding of AI agents

Abstract
This talk will focus on the current directions of the AI Trust and Safety team at Alberta Machine Intelligence Institute (Amii). With the integration of AI tools into many aspects of our day to day lives, it is of utmost importance that we develop a scientific understanding of these tools. In this talk, we will focus on our vision for scientific agentic research at Amii – combining the existing strengths of the University of Alberta in reinforcement learning (RL), with the T&S team's focus on real-world deployments and failures of AI agents. We will then further explore the different ways in which RL could be applicable in this setting – firstly, as a lens through which we examine problems in agentic AI, and secondly as a tool to automate safety testing. We will close with a call to action: bringing together the people building agents with those studying them at a granular level to ensure that agents are deployed responsibly. Trust and safety should be built in from the start, rather than added as an afterthought.

Presenter Bio
Reggie is an Applied Research Scientist on the Trust and Safety team at Amii, where he is leading the development of the agentic AI safety research program. Prior to joining Amii, Reggie completed his Ph.D. at Toronto Metropolitan University, with a research visit hosted by Marlos C. Machado during the last year of Reggie's studies. Reggie's Ph.D. research investigated the manner in which multi-task RL agents could be trained on a wider distribution of tasks than previously thought possible. Reggie completed his MSc at Brock University in swarm intelligence algorithms, was a varsity volleyball athlete, and between his MSc and Ph.D. spent time as the Lead Machine Learning Developer at Castle Ridge Asset Management. During his free time, Reggie enjoys spending time outdoors with his wife and dogs, reading fantasy and science fiction novels, and running.

Website

AI Trust & Safety

Timing & Location

UComm Seminar Room 2-108

pizza from 11:30, seminar from noon to 1

Add event to calendar

Add event

Speaker

Parham Mohammad Panahi, PhD student at the University of Alberta, supervised by Dr. Adam White and Dr. Michael Bowling

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

Dr. Alasdair Newson, Associate Professor at Sorbonne Université, hosted in Edmonton by Dr. Martha White

Title

Generative Models for Image and Video Inpainting and Editing, and Steering of Multimodal Large Language Models

Abstract
In this talk, I will discuss three topics: video inpainting, image editing and steering of Multimodal Large Language Models (MLLMs). The first two subjects employ the impressive capabilities of deep generative neural networks, in particular Diffusion Models and Flow Matching, which I will also briefly introduce.

The first work focuses on video inpainting, that is, filling in missing or damaged regions in videos. We propose an approach based on diffusion models, a family of particularly powerful generative models that rely on the progressive inversion of a noising process. Unfortunately, the large size of these models makes them difficult to use, especially for high-dimensional data such as videos. Therefore, in this work, we propose a ""frugal"" approach to video inpainting, based on the assumption of self-similarity in videos, i.e., the presence of redundant content. We obtain results that are equivalent or superior to state-of-the-art models, while using much more compact models (by one or more orders of magnitude). The associated source code is available here: https://github.com/ncherel/infusion.

In the second part, I will present a series of works which employ generative models for image editing, with the goal of modifying semantic content of images, such as the expression of a face. These methods structure the latent spaces of generative models to correspond to the desired semantic attributes, enabling simple and controllable editing.

Finally, I will discuss the steering of MLLMs. These are models which can take different modalities such as images, text and audio, and produce some output, often text. Unfortunately, this output may be considered dangerous or undesirable, for example giving unfounded financial advice. The task of guiding the internal representation of a fixed MLLM, such that it produce safe outputs, is referred to as ""steering"". In this work, we train a very small Multi-Layer Perceptron to predict an input-dependent steering shift vector. We show that it indeed increases the safety of the responses in popular LLMs such as LLaVa and Qwen2. 

Presenter Bio
Alasdair Newson has been an Associate Professor at the ISIR (Institut des Systèmes Intelligents et de Robotique) of Sorbonne Université since 2023. He received his PhD from Télécom Paris in 2014, under the supervision of Andrés Almansa, Yann Gousseau, and Patrick Pérez. From 2014 to 2018, he was a postdoctoral researcher at Duke University with Guillermo Sapiro, Université Paris Descartes with Bruno Galerne and Julie Delon, and Télécom Paris with Yann Gousseau and Saïd Ladjal. From 2018 to 2023, he was Assistant Professor at Télécom Paris. His research interests include image and video restoration, editing and analysis, and the study of generative models for these purposes.

Website

Alasdair Newson

Timing & Location

UComm Seminar Room 2-108

pizza from 11:30, seminar from noon to 1

Add event to calendar

Add event

Speaker

Matthew Vandergrift, 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

Speaker

TBA

Speaker

TBA

Speaker

Ning Shi, PhD student at the University of Alberta, & David Basil, MSc student at the University of Alberta, supervised by Dr. Greg Kondrak

Title

LLMs and Concepts in Natural Language

Abstract
TBA

Presenter Bios
Ning Shi is a PhD candidate working with Prof. Greg Kondrak. Before starting at the UofA, he worked as a senior algorithm engineer at Alibaba Group. Ning has graduate degrees from Georgia Tech, Syracuse University, and NYU. His doctoral research centers on establishing theoretical and computational connections between various semantic tasks in natural language processing, with the aim of enhancing their understanding and interpretability. 

David Basil is an MSc student working with Prof. Greg Kondrak. His research focuses on cross-lingual lexical semantics, with an emphasis on evaluating conceptual correspondence in translation.

Timing & Location

UComm Seminar Room 2-108

pizza from 11:30, seminar from noon to 1

Add event to calendar

Add event