"The Dimensions of dimensionality" is out in TiCS w @BDRoads We hope to interest those new to embedding approaches and pros. Besides serving as a pracitcal guide, you can pwn your colleagues when they report dimensionality alone as a meaningful result. doi.org/10.1016/j.tics…

The paper discusses how researchers in cognitive science often infer multidimensional representations, called "embeddings," from various types of data, such as text, neuroimaging, neural networks, and human judgments. These embeddings reveal the underlying structure and statistical regularities in the data. The paper highlights that different embedding algorithms prioritize different latent space properties, leading to diverse trade-offs. For example, some algorithms may yield globally interpretable dimensions, while others may produce more compact representations with local interpretability. The dimensionality of the latent space alone does not necessarily indicate the amount of information captured. full paper: openread.academy/en/paper/readi…

@ProfData @BDRoads 🤣 Great motivation to read it (pwnage). It's been an open tab for me for a few days but it does look like a great article and I am looking forward to it!