Dr. Idan Schwartz
Multimodal models, Attention mechanisms, Large language models (LLMs), Generative AI, Explainability, Perceptual bias and plug-and-play control
CV
Dr. Schwartz completed his PhD at the Technion and his post-doctorate at Tel Aviv University. In 2024, he joined the faculty of Bar-Ilan University and established the Multimodal Lab, which focuses on developing "multi-sensory intelligence" capable of understanding and connecting text, image, video, and sound.
Alongside his academic work, he brings extensive industry experience—from developing Microsoft’s Assistant and managing catalog systems at eBay, to his current role as Chief Scientist at Aigency.ai. His research has been published in the world's leading AI conferences, including CVPR, NeurIPS, and AAAI.
Research
Multimodal Learning and Generative Models
When we watch a movie, our brain naturally synchronizes lip movements, facial expressions, sound, and music. We instantly notice if the audio is out of sync, if a movement feels out of context, or if something simply "doesn't feel right." For computers, this integration is far more complex. While modern models can generate impressive visual and textual content, they still struggle to maintain precise connections between different types of information.
Dr. Idan Schwartz researches how to create precise Alignment between different data modalities, enabling machines to understand these underlying connections and produce more accurate, controlled outputs. His research integrates multimodal learning and generative models across text, image, video, and audio, drawing inspiration from human cognition to develop advanced attention and control mechanisms.
Key Research Areas:
Multimodal Learning: Synchronization and alignment across multiple data types.
Generative Models: Advanced techniques for content creation.
Attention & Inference-Time Control: Enhancing precision and reliability in deep models.
Cognition & Deep Learning: Leveraging human-inspired insights for AI architectures.
Research:
Nature Applied-Technological: Developing architectures for generative and multimodal models integrated with industry needs.
Career Path:
Graduates of the lab transition into a variety of high-impact roles:
Industry: AI Researchers and Generative Model Developers.
Academia: Researchers and lecturers in the field of Artificial Intelligence.
Entrepreneurship: Founders of startups in Computer Vision and Generative AI.
Publications
Last Updated Date : 01/08/2026
