Today’s interconnected world presents unique challenges for intelligent systems in processing and integrating diverse data modalities, including text, audio, and visual data. However, traditional closed-world paradigms can fall short when faced with unseen classes and novel scenarios, which frequently emerge in complicated real-world environments. We propose the consideration of open-world learning as a way to build intelligent systems that are highly adaptable while also being robust and trustworthy, capable of tackling highly dynamic and creative tasks. Here, the integration of privacy-preserving techniques is crucial as data sources expand, particularly in high-stakes applications such as autonomous navigation systems for public safety. These systems must discern and adapt to evolving traffic patterns, weather conditions, and user behaviors in real time, underscoring the necessity of continuous learning and resilience against adversities. By exploring these critical challenges, this workshop aims to foster discussions that advance the development of trustworthy, multi-modal systems capable of thriving in open-world contexts.
Universit`a di Modena e Reggio Emilia |
University of California at Merced |
University of Sydney |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Australian National University |
University of Technology Nuremberg |
Princeton University |
Open-World Multi-Modal Learning: Strategies to train systems on both labeled and unlabeled data while distinguishing known from unknown classes.
Dynamic Multi-Modal Vocabulary Expansion: Approaches to enable models to recognize and adapt to an expanding range of concepts using diverse inputs.
Robustness in Multi-Modal Systems: Methods to improve resilience against distribution shifts, adversarial attacks, and noise across modalities.
Multi-Modal Class Discovery: Techniques for identifying new classes across different modalities (e.g., visual, text, audio, touch).
Continual and Federated Learning for Multi-Modal Data: Innovative techniques that support ongoing learning and adaptation in decentralized environments.
Application-Focused Contributions: Research that showcases specific applications, such as autonomous navigation systems in urban environments that leverage multi-modal data for enhanced decision-making.
We will select six oral paper and set one best paper award according to the review results and presentation of a paper.
| Dates and Deadlines | |
|---|---|
| Workshop paper submission | 24 March, 2025 |
| Workshop paper notification | 1 April, 2025 |
| Workshop paper camera - ready | 7 April, 2025 |
| Workshops | 11 June, 2025 |
Date: Wednesday, June 11, 2025, 08:15 AM CDT
Venue: Room 101 E, Music City Center, Nashville, TN
| 8:00 - 8:15 | Opening |
|---|---|
| 8:15 - 8:25 | Oral1: Benchmarking Multi-modal Semantic Segmentation under Sensor Failures: Missing and Noisy Modality Robustness |
| 8:25 - 8:35 | Oral2: IBD: Alleviating Hallucinations in Large Vision-Language Models via Image-Biased Decoding |
| 8:35 - 8:45 | Oral3: A Survey of State of the Art Large Vision Language Models: Benchmark Evaluations and Challenges |
| 8:45 - 9:15 | Invited Talk1: Olga Russakovsky (Princeton University) Title: Defining, Identifying and Generating Out-of-Distribution |
| 9:15 - 9:45 | Invited Talk2: Liang Zheng (Australian National University) Title: Features for Trust-worthy Generative Models |
| 9:45 - 10:00 | Coffee Break |
| 10:00 - 10:30 | Invited Talk3: Lorenzo Baraldi (University of Modena and Reggio Emilia) on behalf of Rita Cucchiara Title: TBD |
| 10:30 - 11:00 | Invited Talk4: Ming-Hsuan Yang (University of California at Merced) Title: Toward Grounding Anything in Images and Videos |
| 11:00 - 12:00 | Poster Session |
| 14:30-14:40 | Oral4: NuanDPO: Nuanced Cross-Modal Preference Optimization for Reducing Hallucinations in Multimodal LLMs |
|---|---|
| 14:40-14:50 | Oral5: Machine Unlearning in Hyperbolic vs. Euclidean Multimodal Contrastive Learning: Adapting Alignment Calibration to MERU |
| 14:50-15:00 | Oral6: Attention-Guided Hierarchical Defense for Multimodal Attacks in Vision-Language Models |
| 15:00-15:30 | Invited Talk5: Yuki M. Asano (University of Technology Nuremberg) Title: Learning from Moving in the World and Generalising from Language Models |
| 15:30-16:00 | Invited Talk6: Tongliang Liu (University of Sydney) Title: TBD |
| 16:00-16:10 | Conclusion |
| 16:10-17:00 | Poster Session |
| Paper No. | Paper Title |
|---|---|
| 1 | Benchmarking Multi-modal Semantic Segmentation under Sensor Failures: Missing and Noisy Modality Robustness |
| 2 | A Survey of State of the Art Large Vision Language Models: Benchmark Evaluations and Challenges |
| 4 | Attention-Guided Hierarchical Defense for Multimodal Attacks in Vision-Language Models |
| 5 | On the Robustness of GUI Grounding Models Against Image Attacks |
| 6 | IBD: Alleviating Hallucinations in Large Vision-Language Models via Image-Biased Decoding |
| 7 | Multimodal Generalized Category Discovery |
| 8 | Machine Unlearning in Hyperbolic vs. Euclidean Multimodal Contrastive Learning: Adapting Alignment Calibration to MERU |
| 13 | PAINT: Paying Attention to INformed Tokens to Mitigate Hallucination in Large Vision-Language Model |
| 14 | NuanDPO: Nuanced Cross-Modal Preference Optimization for Reducing Hallucinations in Multimodal LLMs |
| 15 | HARMONY: Hidden Activation Representations and Model Output-Aware Uncertainty Estimation for Vision-Language Models |
| 17 | Vision Language Models for Massive MIMO Semantic Communication |
University of Trento |
New York University |
National Institute of Informatics |
University of Trento |
Nanjing University |
Osaka University |
Hefei University of Technology |
New York University |
Zhejiang University |
