“Overcoming Output Dimension Collapse: When Sparsity Enables Zero-shot Brain-to-Image Reconstruction at Small Data Scales” by Kenya Otsuka, Yoshihiro Nagano, and Yukiyasu Kamitani has been published in Transactions on Machine Learning Research (TMLR). The paper mathematically clarifies why zero-shot reconstruction of unseen stimuli is possible from a small number of brain–image pairs—not by fully connecting brain features to image features with naive regression, but by assuming sparse feature mappings, building on the framework introduced in Miyawaki et al. (2008). [OpenReview] [arXiv] [Code]