Melbourne RIKEN-AIP
Joint Workshop
School of Computing and Information Systems, The University of Melbourne
This is a joint workshop among AI researchers in both University of Melbourne and RIKEN-AIP, which aims to foster potential collaborations in AI research and AI impact.
Overview
The AI community is developing very fast. Efficient and effective communication is essential for exchanging ideas, building collaborations, and conducting impactful research, etc. This regional workshop aims to bring together some AI researchers to facilitate communication. In this workshop, we propose to showcase a variety of recent advances from presenters, with three emphases, i.e., (1) exchanging ideas; (2) inspiring the juniors to do better research; (3) promoting potential collaborations.
SCHEDULE (Day 1)
Manhari Room, Melbourne Connect, 700 Swanston Street
08:45 - 17:50, Thursday, 3rd July, 2025
14:00 - 15:30
SESSION - Machine Learning and Statistics
Session Chair: Zesheng Ye
In this session, three talks will be presented on machine learning and statistics.
14:00 - 14:30
Invited Talk
Presenter: Ming-Kun Xie (AIP-Imperfect Information Learning Team)
Title: Unlocking the Power of Co-occurrence: A Counterfactual Reasoning Perspective
14:30 - 15:00
Invited Talk
Presenter: Mingming Gong (UoM-MS)
Title: High-Fidelity and Efficient Image Generation via Semi-Implicit Models
15:00 - 15:30
Invited Talk
Presenter: Jean Honorio (UoM-CIS)
Title: Computational and Statistical Foundations in Combinatorial and Non-Convex ML
15:30 - 16:40
Afternoon Tea and Poster Session
Coffee Break and Poster Sessions
Poster 1: Exploring Insights with Partial Labels, Gang Niu
Poster 2: Domain Adaptation and Entanglement: an Optimal Transport Perspective, Okan Koc
Poster 3: Reward Modeling in Reinforcement and Imitation Learning: Overcoming Challenges of Dimensionality, Xin-Qiang Cai
Poster 4: Realistic Evaluation of Deep Partial-Label Learning Algorithms, Wei Wang
Poster 5: Recursive Reward Aggregation, Yuting Tang
Poster 6: Adversarial Multi-view Learning, Haonan Huang
Poster 7: Membership Inference Attack Should Move On to Distributional Statistics for Distilled Generative Models, Muxing Li
Poster 8: S-CPD: Topological Smoothing-based Change Point Detection, Harindu Udage Kankanange
16:40 - 17:50
SESSION - Industry
Session Chair: Zesheng Ye
In this session, two talks will be presented on AI related industry.
16:40 - 17:20
Invited Talk
Presenter: Julien Monteil and Vu Nguyen (Amazon)
Title: Real-time and cold-start recommendation at scale
17:20 - 17:50
Invited Talk
Presenter: Kristen Moore (Data61)
Title: Data61 AI & Cybersecurity: Research Highlights and Future Directions
18:30 - 21:00
Dinner
Location: University House
08:45 - 10:00
SESSION - Opening and Theory
Session Chair: Feng Liu
In this session, the opening remarks will be given for the workshop, and five talks will be presented, including two talks on theory and three talks on AI impact.
08:45 - 09:00
OPENING REMARKS
Presenters: Prof Masashi Sugiyama and Prof Eduard Hovy
Prof Sugiyama and Prof Hovy will give opening remarks for the workshop.
09:00 - 09:30
Invited Talk
Presenter: Masashi Sugiyama (AIP-Imperfect Information Learning Team)
Title: Can We Estimate the Bayes Error Accurately?
09:30 - 10:00
Invited Talk
Presenter: Liam Hodgkinson (UoM-MS)
Title: Heavy-Tailed Universality in Deep Learning
10:00 - 10:30
Morning Tea
Coffee Break
10:30 - 12:00
SESSION - AI Impact
Session Chair: Feng Liu
In this session, three talks will be presented on AI impact.
10:30 - 11:00
Invited Talk
Presenter: Sherief Hashima (AIP-Computational Learning Theory Team)
Title: AI-Driven RIS: Unleashing the Full Potential of 6G Networks
11:00 - 11:30
Invited Talk
Presenter: Eun Jung (EJ) Holden (UoM-CIS)
Title: Geospatial Information Retrieval for Subsurface Modelling from Industry Perspective
11:30 - 12:00
Invited Talk
Presenter: Van-Quang Nguyen (AIP-Robotics for Infrastructure Management Team)
Title: Multimodal LLM Applications for Domain-specific Tasks
12:00 - 14:00
Lunch
Location: M01+M02, Level M, Melbourne Connect
13:00 - 15:00
SESSION - NLP, Vision, and Multi-modalities
Session Chair: Gang Niu
In this session, four talks will be presented on NLP, Vision, and Multi-modalities.
13:00 - 13:30
Invited Talk
Presenter: Benjamin Heinzerling (AIP-Natural Language Understanding Team)
Title: Analyzing and Interpreting the Internals of LLMs
13:30 - 14:00
Invited Talk
Presenter: Kris Ehinger (UoM-CIS)
Title: Open-World Scene Perception and Reasoning with Diffusion Models and MLLMs
14:00 - 14:30
Invited Talk
Presenter: Nir Lipovetzky (UoM-CIS)
Title: Why Planning is Hard for LLMs? The Lens of Planning for Novelty over Common Problems in Control, Planning and RL
14:30 - 15:00
Invited Talk
Presenter: Eduard Hovy (UoM-CIS)
Title: Toward Mechanistic Interpretation of LLM Inference
15:00 - 15:30
Afternoon Tea
Coffee Break
15:30 - 17:00
SESSION - NLP, Vision, and Multi-modalities
Session Chair: Gang Niu
In this session, four talks will be presented on NLP, Vision, and Multi-modalities.
15:30 - 16:00
Invited Talk
Presenter: Christine de Kock (UoM-CIS)
Title: Forecasting the socio-linguistic evolution of online extremist communities
16:00 - 16:30
Invited Talk
Presenter: Ting Dang (UoM-CIS)
Title: Human-Centred AI for Mobile Health Sensing and Computing
16:30 - 17:00
Invited Talk
Presenter: Yuning Qiu (AIP-Tensor Learning Team)
Title: STEPS: Sequential Probability Tensor Estimation for Text-to-Image Hard Prompt Search
17:00 - 17:30
Invited Talk
Presenter: Mingyuan Bai (AIP-Tensor Learning Team)
Title: Foundation Models for Adversarial Defense
17:30 - 18:00
Networking/Working
Researchers can chat with each other on interesting topics or work together on the white boards.
Closing remarks
Feng will conclude the workshop.
09:00 - 10:00
SESSION - Machine Learning and Statistics
Session Chair: Zesheng Ye
In this session, three talks will be presented on machine learning and statistics.
09:00 - 09:30
Invited Talk
Presenter: Eshant English (AIP-High-Dimensional Causal Analysis Team)
Title: Minimax Optimal Convergence for Stochastic Interpolants
09:30 - 10:00
Invited Talk
Presenter: Hanxun Huang (UoM-CIS)
Title: Understanding Adversarial and Backdoor Vulnerabilities in CLIP
10:00 - 10:30
Morning Tea
Coffee Break
10:30 - 12:00
SESSION - Machine Learning and Statistics
Session Chair: Zesheng Ye
In this session, three talks will be presented on machine learning and statistics.
10:30 - 11:00
Invited Talk
Presenter: Zhen-Yu Zhang (AIP-Imperfect Information Learning Team)
Title: In-context Demonstration Matters: On Prompt Optimization for Pseudo-Supervision Refinement
11:00 - 11:30
Invited Talk
Presenter: Susan Wei (Monash)
Title: Transformers as Bayesian Predictive Engines
11:30 - 12:00
Invited Talk
Presenter: Wei Huang (AIP-Deep Learning Theory Team)
Title: Understanding the Optimization and Generalization of Transformers
12:00 - 13:00
Lunch
Location: Manhari Room
SCHEDULE (Day 2)
Manhari Room, Melbourne Connect, 700 Swanston Street
09:00 - 18:00, Friday, 4th July, 2025
ORGANISERS
This workshop is co-organised by the Melbourne TMLR group and RIKEN-AIP.
Program Co-chairs:
-
Feng Liu, University of Melbourne
-
Zesheng Ye, University of Melbourne
-
Gang Niu, RIKEN-AIP
-
Masashi Sugiyama, RIKEN-AIP
Publicity and Workflow Co-chairs:
-
Ruijiang Dong, University of Melbourne
-
Muxing Li, University of Melbourne
​
We also thank our excellent volunteers for taking care of AIP guests:
-
Yiyi Guo, University of Melbourne
-
Jinhao Li, University of Melbourne
-
Haoyuan Li, University of Melbourne
-
Zhijian Zhou, University of Melbourne
SPONSORS
GET IN TOUCH
If you have questions about the seminars, don’t hesitate to reach out.
