Georgia Institute of Technology

Binary protein classification over frozen ESM3 embeddings, using transformer contextualization and learned attention pooling instead of fixed pooling.

Intracortical speech brain-to-text decoder jointly pretrained across six BCI users, with over 50% lower relative word error than per-user models.

Self-supervised brain MRI foundation model built on DINOv3, pretrained on roughly 6.6 million unlabeled axial slices for neuroimaging tasks.

Cell-centric microscopy foundation model that distills morphology and microenvironment views into a unified embedding for virtual spatial omics.

Vision Transformer foundation model for spatial metabolomics, pretrained on ~4,000 curated METASPACE mass spectrometry imaging datasets.

Hierarchical transformer for virus discovery in metagenomes, classifying viral genomes across taxonomic ranks and flagging candidate novel lineages.

Protein foundation model built on Bayesian Flow Networks, prompted with MSA profiles for structure- and function-preserving sequence design.

Medical vision-language model trained with reinforcement learning for generalizable reasoning across eight imaging modalities and five question types.

Protein function annotation that reshapes language model embeddings with a neural-collapse loss so rare EC, Pfam, and GO classes stay separable.

EEG foundation model pretrained by momentum contrastive learning and masked reconstruction, with a learnable channel mapping that unifies montages.

Molecular docking framework that poses several ligands sharing one protein pocket at once, using their consistency to sharpen each prediction.

Conditional diffusion model with cross-attention that synthesizes subject-specific 3D intrinsic connectivity networks from resting-state fMRI.

Spatiotemporal vision transformer that turns a resting-state fMRI scan into 4D brain network maps, supervised by windowed ICA components.

Multimodal RNA chat model pairing a frozen RNA-FM encoder with a Llama-3 8B backbone to answer natural-language questions about a submitted sequence.

Foundation model for photoplethysmography (PPG) that learns quality-robust waveform representations for heart rate, blood pressure, and AF detection.