🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

PyTorch Geometric Signed Directed is out! Head over to our paper : arxiv.org/abs/2202.10793, GitHub: github.com/SherylHYX/pyto… and documentation: …metric-signed-directed.readthedocs.io/en/latest/ to find out more! Thanks to Xitong Zhang , Junjie Huang, Mihai Cucuringu and Gesine Reinert . #gnn #datascience

Implicit gradient regularization by @dgtbarret and عبد امل showed that gradient descent trajectory is closer to a modified flow (than gradient flow). This modified flow minimizes the loss E and the norm of the gradient h/4||\grad E||^2, preferring shallower paths. (h: step size)