Events

While Large Language Models (LLMs) exhibit strong zero-shot capabilities, carefully designed inference-time strategies are crucial for unlocking their full potential. This talk delves into two tasks where this is particularly evident: text ranking and retrieval-augmented generation (RAG).

This thesis proposes several key works as solutions in decentralized POI recommender systems.

The central focus of our research: developing large language models and foundational models specifically tailored for graph data.

In this talk, I will outline the progress in LLM research, and the opportunities they afford to multilingual settings.

In this talk, I will discuss an alternative approach through inverse rendering, which enables machine learning models to extract explicit physical representations from raw, unstructured image data, such as Internet photos and videos.

This presentation embarks on a comprehensive exploration of the VLN trajectory, tracing its inception to seminal benchmarks such as Room-to-Room (R2R).

This project aims to develop a multi-modal, multitask diagnostic algorithm that leverages several novel deep learning innovations.

The EECS EMCR Advocacy Committee invites you to a panel session themed around entrepreneurship development.

This thesis focuses on practical scenarios of domain adaptation.

In this talk, we will introduce: (1) The long-term technical goal will the GFMs serve (2) The knowledge gap in the graph domain the GFMs can fill (3) The critical problem GFMs can solve.

Searching the gigantic corpus of online podcasts involves multiple challenges ranging from content and style diversity to expensive audio processing to variable length. In this thesis, we aim to address these challenges and devise novel approaches to improve state-of-the-art performance.

Our primary research question is how we can effectively utilize limited labelled data for irregular time series modelling?

In this presentation, we focus on expediting ViTs while maintaining high performance.

This presentation proposes a series of compression approaches to reduce the computation complexity and memory usage of deep models for efficiency improvement.

Entity Alignment (EA) is crucial for integrating heterogeneous knowledge graphs (KGs) into a unified knowledge base by identifying equivalent entities across them.

This talk explores the significance of personalization and its potential impact on mood improvement through AI-Composed Soundscapes.

This presentation explores the comparative efficacy and influencing factors of three major platforms: Search Engines, Symptom Checkers, and Large Language Model (LLM)-powered Conversational Agents.

Extensive experiments on public benchmark datasets have verified the effectiveness of the two proposed works in retaining excellent recommendation performance.

To ensure DNNs effectively retain past knowledge while accommodating future tasks, we explore CL techniques from the viewpoint of augmenting and modularizing the memorization of DNNs.

In this work, we use MetaMap to map the original dataset, which lacks ground truth, with knowledge from drug-drug interaction databases, thereby deriving the ground truth dataset.

This presentation shows results both on synthetic data and real network+smart meter data from a Belgian system operator.

Experience new ways to interact with digital technology in this rapidly evolving field at the 2024 UQ Interaction Design Exhibit.

In this talk, we introduce the Sequential Recommendation problem and draw parallels between language modelling and recommender systems.

Plant disease recognition aims to identify disease through visual symptoms, and has a profound impact on agricultural production and food safety.

In this talk, we will discuss the key challenges in improving efficiency of LLM serving.