Paper accepted by TMLR.

📣 📣 📣

One of my co-authored papers, AI-SARAH: Adaptive and Implicit Stochastic Recursive Gradient Methods was accepted by Transactions of Machine Learing Research (TMLR).

AI-SARAH is the first plug-n-play algorithm in the realm of stochastic recursive gradient method. It automatically adapts to ML models and can save hundreds, if not thousands, of experiments performing hyper-parameter fine-tuning. It is super easy and efficient to implement in practice, and it outperforms many state-of-the-art methods, including ADAM and SGD, in solving machine learning problems.