Events

On behalf of .au Domain Administration, the School of Electrical Engineering and Computer Science invites UQ students and staff to attend a presentation on the multi-stakeholder model of internet governance.

This study delves into detection, evaluation, and prediction tasks within the context of OOD on tabular datasets.

This CPD course will feature presentations by Prof. Dr.- Ing. Stefan Tenbohlen from the Institute of Power Transmission and High Voltage Technology at the University of Stuttgart in Germany. This course will deliver the theoretical background information necessary to develop a good understanding of the phenomena of partial discharge (PD) in high-voltage equipment, as well as current measurement techniques for partial discharge. Along with this, the course will look at the experience of industry using partial discharge for diagnostic and testing work. Using the high-voltage laboratory at the University of Queensland, there will be demonstrations of some key measurement techniques for partial discharge.

Towards Open-Vocabulary 3D Object Detection.

Despite the advancements of DecRecs, they face unique challenges, including constrained local storage, inflexible architectures, vulnerability to poisoning attacks, and performance degradation.

This research highlights the complex security challenges in FedRecs arising from their open and decentralized nature.

We propose a novel framework named Memory-enhanced Invariant Prompt learning (MIP) for urban flow prediction under constant distribution shifts.

As the EMCR Informal Talk series are designed to share peoples’ own lives, keeping the Informal Talk as an in-person only event best aligns with its personal nature and can maximize interactions with the audience.

This talk explores compositional visual representations bridging pre-trained language and visual generative models through programs for scene structure, words for semantic abstraction, and neural embeddings capturing visual identity.

In this work, we show that optimal ranking performance can be achieved without requiring the full-length embeddings, which leads to improved computational efficiency.

This talk will cover our recent work on scalable hybrid approaches that combine LLMs and traditional recommendation models.

In this talk, I will share our journey behind developing an effective long-context language model.

In this talk, I will first introduce our recent work to fine-tune a pre-trained VLM with downstream tasks, and further improve the generalisation ability of a pre-trained VLM. Then, I will introduce another work on how to detect label set shift when using pre-trained VLMs in zero-shot classification.

Content-based recommender systems (CRSs) play a crucial role in assisting users to navigate the vast information available on the Internet.

This two-day face-to-face course will bring industry professionals together for dialogue and sharing of knowledge to better understand the operation of bushings for power transformers, as well as the design, maintenance and testing thereof.

In this talk, we will introduce recent advances that develop theoretically principled and practical useful methods for learning on graphs in the challenging open-world hypothesis, where the model needs to generalize to out-of-distribution testing data in the wild.

Don't miss the 2024 Innovation Showcase, a unique opportunity for EECS students to showcase their end of year projects to industry partners and the UQ community.

This talk will cover evaluating and training LMs to reason in real-world scenarios.

In this talk, I will introduce Linear-Time Graph Neural Network (LTGNN), a breakthrough approach designed to bridge the scalability gap between GNNs and simpler methods without sacrificing GNNs’ unique ability to capture distant dependencies.

This two-day face-to-face course will bring industry professionals together for dialogue and knowledge sharing to better understand the renewable energy technologies and their integration regarding renewable generator modelling, control techniques, frequency and voltage regulation aligning with grid codes.

This two-day face-to-face course will bring industry professionals together for dialogue and sharing of knowledge to better understand the fundamentals of power systems along with its modelling and operational aspects.

In this work, we argue that dense document embeddings, while effective, are implicitly out-of-context for targeted use cases of retrieval, and that a contextualized document embedding should take into account both the document and neighboring documents in context - analogous to contextualized word embeddings.

This talk offers a comprehensive review of GNN applications in epidemic modeling, presenting a hierarchical taxonomy for both epidemiological tasks and modeling techniques.

We introduce Dispersing Prompt Expansion DiPEx, a self-supervised prompt learning strategy that overcomes the limitations of manually crafted text queries in VLMs, which often miss objects due to semantic overlap diminishing detection confidence.

This PhD study aims to tackle the common challenges by designing novel training strategies for deep learning models trained on imperfect datasets.