Meta-Learning-Papers

Why we included this project

Meta-learning research sits scattered across two decades of conference papers, and this list does the sorting for you. It collects several hundred papers on learning to learn, one-shot learning, and few-shot learning, grouped by technique family, so you can follow how the field moved from early ideas like Schmidhuber's embedded meta-levels and Hochreiter's learning-to-learn gradient methods to the gradient-based, metric-learning, and model-agnostic work that dominates today. Graduate students and researchers starting out get a sensible reading order without having to guess which papers matter. Practitioners comparing few-shot algorithms can jump straight to the technique they care about and find the seminal work behind current methods. It is a curated bibliography, not software; expect paper lists and venues, not code you can run.