Language and Multimodal AI Lab (LAMA)
The Language and Multimodal AI Lab (LAMA)
aims to develop machine learning algorithms for language understanding and generation following a holistic approach where language can be combined with other types of information. In addition to textual context of varying length, our models include knowledge in the form of images, videos, audio recordings, sensor data, metadata and knowledge bases. Multimodal machine learning is therefore at the core of our research. The main areas we work on include (multimodal) language generation, machine translation, text adaptation, image captioning, transfer learning for language, NLP for health and sustainable development, quality evaluation and estimation, and language learning applications.