Fabian Schmich

scifAI provides an open-source framework for analyzing imaging flow cytometry data to predict antibody function and characterize immunological synapses.

PXPermute identifies important fluorescent channels in IFC data by permuting pixel values and evaluating model performance, showing superior accuracy in channel selection.

The paper develops a model to infer pairwise epistatic gene interactions from siRNA perturbation screens, showing that glinternet outperforms xyz in accuracy for high-dimensional data.

The paper compares incomplete-supervision methods for biomedical image classification, finding that pre-training and semi-supervised learning contribute more than active learning to performance.

The paper improves pathway reconstruction from RNAi screens by incorporating off-target effects into nested effects models, enhancing network accuracy.