Postdoctoral position on multimodal and generative AI models for scene understanding
The Artificial Intelligence for Good (AIGO) Research Unit is coordinated by Prof. Vittorio Murino. It focuses on fundamental AI topics from methodological and theoretical perspectives, yet functional to tackle a number of applications and actual case studies related to several domains such as biomedical, healthcare, and several others ultimately leading to people wellbeing.
Specifically, AIGO aims at studying learning paradigms in presence of imperfect data, especially in multimodal scenarios, hence tackling unsupervised, semi-supervised and self-supervised settings, weakly or noisy labeled data, few, class imbalanced, or biased. Domain adaptation and generalization, few/zero-shot learning, learning with biased data, and continual learning, even extended to multimodal scenarios, are among the major areas to be investigated, also given their valence in tackling practical, real-world applications.
AIGO will also consider generative AI models, especially related to the most recent trend regarding multimodal foundation models, including large language models (LLMs) and vision and language models (VLMs). From a general standpoint, AIGO will also consider architectural and computational issues addressing the so-called lightweight Machine Learning in order to design low-power ML technologies and approaches dealing with machine intelligence at the very edge of the cloud (e.g., for robots). Given the focus on people wellbeing, we clearly aim at the design of AI methods considering at their ground ethical, privacy and fairness aspects, as well as their robustness, towards the design of explainable, trustworthy and transparent deep learning techniques.
Main applications will involve biomedical, biological, neuroscience, and healthcare in general. Brain investigation (and its diseases) is identified as the main (but not exclusive) area of interest. Ultimately, AIGO will seek to develop models that can also be readily applicable to IIT interdisciplinary research, ranging from neuroscience to robotics, in particular by leveraging our in-house robotics platforms (iCub, ErgoCub, R1 et al.), IIT neuroscience teams, and HPC computational facilities.
