KolTa active learning framework discovers top formulations with limited queries
KolTa selects batches for top 5% formulations using PRC-UCB, outperforming comparators on viscosity, solubility, and conductivity tasks.
Stories per day · 30 days
2 RSS
Relatedactive-learning1formulation-discovery1gpu1korteweg-de-vries1neural-emulation1polynets1training-free1
6 stories, newest first · one card per story
Stories tagged #physics-informed
The Physics Is the Network: Fast, Training-Free Neural Emulation
Training-free PolyNets emulate known polynomial physics with solver-level accuracy and strong GPU ensemble scaling.
Learning Traffic Flow Dynamics with Stochastic Physics-Informed Neural Cellular Automata
A physics-informed neural cellular automaton learns stochastic traffic rules while conserving vehicles and respecting road topology.
SoftServe: A Scalable Quasi-Newton Method for Deep Learning
SoftServe scales positive-definite quasi-Newton updates to large networks and beats Adam, Muon, and SOAP on ill-conditioned tasks.
Port-Hamiltonian Neural Networks for Systems with Multiple Asymptotically Stable Equilibria
Port-Hamiltonian networks are extended to multiple stable equilibria using Bregman divergences.
GridSFM: A Foundation Model for Solving AC Optimal Power Flow
GridSFM, a 15-million-parameter graph model, solves AC optimal power flow on grids up to 10,000 buses.