Advancing Large AI Models: Integration of New Data Modalities

This grant aims to support research institutions and AI development teams in the European Union by advancing the capabilities of large AI systems. This initiative focuses on integrating innovative data modalities into AI models and improving the performance and application potential of multimodal AI systems.

Type of Project to Develop

• Projects that integrate new types of data, such as sensor measurements or remote sensing, into large AI models.
• Projects that enhance the capabilities of current multimodal AI systems to handle more types of data and perform better in various tasks.
• Projects that develop new methodologies for the pretraining and fine-tuning of multimodal AI models to improve robustness, efficiency, and trustworthiness.

Eligibility Criteria

• Research institutions and organizations within the EU.
• Teams with expertise in AI, machine learning, and data integration.
• Proposals that address multidisciplinary aspects, including ethics and bias mitigation.

Eligible Expenses

• Costs for data collection, processing, and cross-modal alignment.
• Costs for development and testing of AI models, including pretraining and fine-tuning activities.
• Costs for computing resources, evaluation procedures, and dissemination of results.

Examples of Eligible Projects

• A research project integrating real-time sensor data from IoT devices into AI models for improved predictive maintenance in manufacturing.
• An initiative developing a multimodal AI system that combines text, images, and speech data to enhance multilingual virtual assistants.
• A project creating advanced AI models for weather forecasting by incorporating remote sensing data and improving the models' accuracy and reliability.