An exploration of how to use AI-generated human faces to visualize high-dimensional data (sort of) via an eigenspace projection

An extremely high-level overview of the ways we regularize neural networks, their pros and cons, and where we might go next

Using knowledge of distinct classes within a training distribution to structure an autoencoder-learned latent state

Randomly shuffling sequences (as strings or one-hot-encoded NumPy arrays) to preserve k-mer frequencies

Vectorizing computation of Spearman correlation and auPRC using NumPy