ACM MM 2024深度多模态生成和检索研讨会，诚邀各界专家学者参与

最新推荐文章于 2024-08-11 22:15:43 发布

PaperWeekly

最新推荐文章于 2024-08-11 22:15:43 发布

阅读量294

点赞数

原文链接：https://mp.weixin.qq.com/s?__biz=MzIwMTc4ODE0Mw==&mid=2247665839&idx=4&sn=fc099ab6b8fc9aace6fec4c165dd60a0&chksm=97d899ef1816a2d2a40823ca67be4edea8cfcda96ce55e1c0f15704956b2077b9c2daf7aecb5&scene=126&sessionid=0

版权

简介

The 2nd International Workshop on Deep Multimodal Generation and Retrieval (MMGR) at ACM Multimedia 2024 focuses on the advancements in deep multimodal learning, emphasizing the integration of diverse data modalities such as text, images, audio, and video. This workshop aims to bring together researchers and practitioners to discuss original ideas, perspectives, and challenges in multimodal semantics understanding, generative models, multimodal information retrieval, and explainable and reliable multimodal learning.

信息生成（IG）和信息检索（IR）是两种主要的信息获取方式，即通过生成或检索来获取内容。虽然传统的IG和IR技术在语言处理领域取得了巨大成功，但在不同模态（如文本、图像、音频和视频）数据源的利用不足，可能会限制IG和IR技术的全面发展，从而影响其在现实世界中的应用。鉴于我们的世界充满了多媒体信息，本次研讨会鼓励在IG和IR研究中发展深度多模态学习。

多模态数据类型和技术，如DALL-E、Stable Diffusion、GPT-4和Sora，展示了在多模态IG和IR学习中的巨大潜力。然而，这些领域中仍存在许多未解决的挑战和开放性问题。通过本次研讨会，我们希望鼓励在深度多模态生成与检索领域的更多探索，为相关方向的研究人员提供一个分享见解和进展的平台，以推动这一快速发展的领域。

重要时间点

Paper Submission

July 19, 2024 (AoE)

Notification of Acceptance

August 5, 2024 (AoE)

Camera-ready Submission

August 19, 2024 (AoE) [Firm Deadline]

Workshop Dates

28 October - 1 November, 2024 (AoE)

论文主题

1. Multimodal Semantics Understanding:

Vision-Language Alignment Analysis
Multimodal Fusion and Embeddings
Large-scale Vision-Language Pre-training
Structured Vision-Language Learning
Commonsense-aware Vision-Language Learning
Visually Grounded Language Parsing

2. Generative Models for Image/Video Synthesis:

Text-free/conditioned Image Synthesis
Temporal Coherence in Video Generation
Image/Video Editing and Inpainting
Visual Style Transfer
Multimodal Dialogue Response Generation

3. Multimodal Information Retrieval:

Image/Video-Text Compositional Retrieval
Image/Video Moment Retrieval
Image/Video Captioning
Multimodal Retrieval with MLLMs

4. Explainable and Reliable Multimodal Learning:

Explainable Multimodal Retrieval

Adversarial Attack and Defense

Multimodal Learning for Social Good

Efficient Learning of MLLMs

提交类型

1. Position or Perspective Papers: Original ideas, perspectives, research visions, and open challenges in the workshop topics. (4 pages + 1-page reference or 8 pages + 2-page reference)

2. Featured Papers: Already published papers or papers summarizing existing publications relevant to the workshop topics.

3. Demonstration Papers: Original or already published prototypes and operational evaluation approaches. (Up to 2 pages + unlimited references)

提交方式

Submissions should be written in English and formatted according to the current ACM two-column conference format. Authors must anonymize their submissions. Suitable LaTeX, Word, and Overleaf templates are available on the ACM website. Submissions can be made through OpenReview.

审稿过程

All submissions will be peer-reviewed by at least two experts in the field. The review process will be two-way anonymized. Accepted papers will be published in the ACM Digital Library, and high-quality papers can be recommended to a special issue in ACM ToMM.

组织者团队

Wei Ji: National University of Singapore

Hao Fei: National University of Singapore

Yinwei Wei: Monash University

Zhedong Zheng: University of Macau

Juncheng Li: National University of Singapore

Long Chen: The Hong Kong University of Science and Technology

Lizi Liao: Singapore Management University

Yueting Zhuang: Zhejiang University

Roger Zimmermann: National University of Singapore