Science & medicine for advanced learners
189 resources · page 6 of 6
GraphCast paper
by Remi Lam et al. (Google DeepMind)
arXiv paper (2022) introducing GraphCast, a graph neural network for medium-range global weather forecasting.
Grid2Op
Grid2Op documentation: Python framework for developing and testing control agents on simulated power grids.
Intern-S1: A Scientific Multimodal Foundation Model
by Lei Bai et al. (Shanghai AI Laboratory)
arXiv PDF (2025) of the Intern-S1 technical report, Shanghai AI Laboratory's scientific multimodal model.
KAN: Kolmogorov-Arnold Networks
arXiv paper (2024) introducing Kolmogorov-Arnold Networks (KAN) with learnable activation functions on edges.
Kornia
Kornia documentation: PyTorch library of differentiable computer-vision and image-processing operations.
Learning to Simulate Complex Physics with Graph Networks (GNS)
by Alvaro Sanchez-Gonzalez et al. (Google)
arXiv paper (2020) introducing Graph Network-based Simulators (GNS) for learning complex particle physics.
MatGL
by Materials Virtual Lab (UC San Diego)
MatGL: graph deep learning library for materials property prediction and interatomic potentials.
matminer
by Hacking Materials group (Lawrence Berkeley National Laboratory)
matminer documentation: Python library for mining materials data and generating features for ML models.
MaxProof
by Jiacheng Chen et al. (Chinese University of Hong Kong)
arXiv paper (2026) introducing MaxProof, generative-verifier RL and test-time scaling for mathematical proofs.
MedHELM: Holistic Evaluation of Large Language Models for Medical Tasks
by Suhana Bedi et al. (Stanford University)
arXiv PDF (2025) of the MedHELM paper on holistic evaluation of LLMs for medical tasks.
MeshGraphNets
by Tobias Pfaff et al. (Google)
arXiv paper (2020) introducing MeshGraphNets for learning mesh-based physical simulation.
MoveIt
MoveIt 2 documentation: ROS 2 framework for robot motion planning, manipulation and perception, with tutorials.
Neural Operator: Learning maps between function spaces
by Nikola Kovachki et al. (Caltech)
arXiv paper (2021) "Neural Operator: Learning Maps Between Function Spaces", the general framework covering Fourier and graph neural operators for PDEs.
NeuralHydrology
NeuralHydrology site: research blog and library for neural networks in rainfall-runoff and hydrological modelling.
NVIDIA cuOpt
by NVIDIA
NVIDIA cuOpt user guide: GPU-accelerated solver for routing and other optimisation problems.
Orfeo ToolBox Cookbook
Orfeo ToolBox Cookbook: CNES open-source remote-sensing library with image classification workflows.
PaiNN
by Kristof T. Schütt et al. (Technische Universität Berlin)
arXiv HTML version of the PaiNN paper (2021) on equivariant message passing for molecular property prediction.
Accurate medium-range global weather forecasting with 3D neural networks (Pangu-Weather)
by Kaifeng Bi et al. (Huawei Cloud)
Nature paper (2023) on Huawei's Pangu-Weather, medium-range forecasting with 3D neural networks.
Physics-informed deep learning (Raissi et al.)
by Maziar Raissi et al. (Brown University)
arXiv paper (2017) introducing physics-informed neural networks (PINNs) for data-driven solutions of nonlinear PDEs.
PX4 computer-vision and ROS integration
by PX4 Autopilot (Dronecode Foundation)
PX4 Autopilot user guide: open-source drone flight control software with simulation, ROS and computer-vision integration.
Raster Vision
Raster Vision: open-source library for deep learning on satellite and aerial imagery.
robomimic
by robomimic
robomimic site: framework and demonstration datasets for robot learning from demonstration.
Scanpy documentation
Scanpy documentation: Python toolkit for analysing single-cell gene expression data.
SchNet: A continuous-filter convolutional neural network for modeling quantum interactions
by Kristof T. Schütt et al. (Technische Universität Berlin)
arXiv paper (2017) introducing SchNet, a continuous-filter convolutional network for modelling quantum interactions in molecules.
SchNet: a deep learning architecture for molecules and materials
by Kristof T. Schütt et al. (Technische Universität Berlin)
arXiv paper (2017) "SchNet - a deep learning architecture for molecules and materials", using continuous-filter convolutions to predict molecular energies and forces.
SUMO
by German Aerospace Center (DLR)
SUMO documentation: open-source microscopic traffic simulator from DLR, used in traffic-control research.
Tokamak plasma control with deep RL
by Jonas Degrave et al. (Google DeepMind)
Nature paper (2022) on magnetic control of tokamak plasmas with deep reinforcement learning.
TorchDrug
TorchDrug: PyTorch library with datasets, models and tutorials for machine learning in drug discovery.
TorchGeo documentation
TorchGeo documentation: PyTorch library of geospatial datasets, samplers, transforms and pretrained models.