Join Us
Join Us
Machine Intelligence lab (MiLab) welcomes applications for Ph.D., Master’s, and undergraduate research positions in the following fields.
[EEG/fMRI Foundation Models]
While recent AI breakthroughs have been driven by foundation models that learn general-purpose representations from massive datasets, EEG- and fMRI-based models remain largely task-specific. Adapting them to new tasks typically requires extensive retraining, architectural modifications, or new labeled data. To address this, EEG/fMRI foundation models seek to learn transferable representations adaptable to various downstream applications. The ultimate goal is to build a shared backbone that generalizes across subjects, devices, and tasks, shifting neuroimaging AI toward a general-purpose paradigm akin to large language models.
[EEG/fMRI Brain Decoding]
Recent AI breakthroughs are revolutionizing our understanding of the human brain and the development of brain–computer interfaces (BCIs). At the forefront of this shift is the application of generative AI to decode neural signals, translating human thoughts and perceptions into text or visual representations. By leveraging the complementary advantages of different neuroimaging modalities, researchers utilize fMRI for deep mechanistic insights and EEG for real-world, practical BCIs. Crucially, emerging EEG-to-fMRI mapping techniques aim to synergize these modalities, enabling highly informative yet broadly accessible brain decoding systems.
[Causal Deep Learning]
While contemporary AI models excel at identifying data patterns, they struggle with causal and counterfactual queries, such as determining disease etiologies or predicting alternative treatment outcomes. Addressing these limitations requires a paradigm shift from mere correlation to genuine cause-and-effect reasoning. To this end, causal discovery seeks to elucidate the underlying causal structures from observational data, while causal inference modeling estimates the effects of interventions, even in the presence of unobserved factors. Increasingly, these frameworks are being integrated with deep learning, yielding causal deep learning approaches capable of capturing complex, nonlinear dynamics in high-dimensional data like images and time series. In the meantime, causal reasoning has emerged as a critical frontier for large language models (LLMs), prompting researchers to investigate whether these models can perform true causal deduction rather than merely replicating statistical patterns.
[AI for Science]
The ‘AI for Science’ paradigm aims to develop next-generation artificial intelligence models capable of operating as domain experts to address intricate biological and chemical challenges. Transcending traditional trial-and-error methodologies, this approach facilitates the autonomous exploration and computation of vast chemobiological spaces that far exceed human capacity. By elucidating the fundamental principles governing molecular structures, biological systems, and disease mechanisms, this research establishes robust computational frameworks. Ultimately, this endeavor seeks to accelerate the entire spectrum of research across chemistry, biology, and novel drug discovery, translating raw scientific data into actionable therapeutic insights.