Multi-modal learning is an innovative field in machine learning that focuses on integrating and leveraging data from multiple diverse sources or modalities. This approach aims to create more robust, accurate, and comprehensive models by combining information from different types of data, such as images, text, audio, and numerical data.

In this summer school, we will explore multi-modal learning with a particular emphasis on integrating images data with other modalities. En studerende, der fuldt ud har opfyldt kursets mål, vil kunne:

• Describe the concept of multi-modal learning
• Describe different scenarios where multi-modal data are acquired
• Import and visualize large multi-modal data using Python
• Do basic manipulation of image data
• Implement and evaluate a deep learning based multi-modal framework
• Use and explain different metrics for evaluating the performance of multi-modal learning frameworks
• Create a short and informative presentation of the main results of an evaluated multi-model learning framework
• Describe state-of-the-art in multi-model learning

Multi-modal learning is an innovative field in machine learning that focuses on integrating and leveraging data from multiple diverse sources or modalities. This approach aims to create more robust, accurate, and comprehensive models by combining information from different types of data, such as images, text, audio, and numerical data.

In this summer school, we will explore multi-modal learning with a particular emphasis on integrating images data with other modalities. We have invited a group of speakers that are specialist in learning from multiple data sources. The course will examine real-world applications across various disciplines, including healthcare (integrating medical scans with patient records) and biology (combining genetic information with visual data of insects), as well as theoretical developments in combining images with e.g. text or video.

By learning to harness the power of multiple data modalities, researchers can generate new insights and tackle complex problems that single-modality approaches may struggle to solve. This course will provide participants with the theoretical foundations and practical skills needed to apply multi-modal learning techniques in their own research domains.

A key component of the summer school will be hands-on, group-based project work. Participants will engage in a programming challenge that applies multi-modal learning techniques to a real-world problem. The format of a challenge will encourage collaboration, creativity, and critical thinking in the practical application of multi-modal learning concepts. https:/​/​multi-modal.compute.dtu.dk/​talks-materials-and-exercises/​ 25. april, 2025