The paper investigates the effectiveness of Δ-learning in scientific machine learning, showing that simply reducing residual error magnitude does not guarantee easier learning. By testing molecular graph neural networks on total energy predictions, the authors find that complex local descriptor baselines can produce small residuals that are actually rougher and harder to learn, whereas a semi‑empirical baseline both shrinks the residual scale and smooths the target space. They propose a new diagnostic, scale‑normalized graph Dirichlet roughness (SD_{R}), to assess residual learnability and argue that choosing complementary baselines is as important as model architecture for successful target design.
arXiv:2606. 03232v1 Announce Type: cross Abstract: Graph Neural Networks (GNNs) have revolutionized Neural Force Fields for atomistic simulations, achieving near-quantum accuracy at reduced cost, yet adapting these models to new chemical systems requires expensive retraining of foundation models.
arXiv:2509. 21624v3 Announce Type: replace Abstract: Fundamental tasks in computational chemistry, from transition state search to vibrational analysis, rely on molecular Hessians, which are the second derivatives of the potential energy.
The paper proposes a three‑stage training pipeline that begins with procedural pretraining on abstract, procedurally generated data, followed by molecular pretraining on SMILES, and finally downstream fine‑tuning for molecular property prediction. Experiments show that procedural pretraining improves downstream performance—e.g., a 4.8% error reduction on Lipophilicity—especially when labeled data are scarce, and that the benefit peaks at an intermediate procedural training budget. Analysis indicates that transferable knowledge resides mainly in attention layers, while feed‑forward layers may over‑specialize.
arXiv:2605. 15511v2 Announce Type: replace Abstract: Graph Neural Networks (GNNs) have become the dominant framework for inductive graph-level learning.
The paper investigates whether machine learning models can uncover physical patterns in atomistic data without relying on traditional physics-based inductive biases such as geometric locality or graph structures. By training a general-purpose architecture on molecular simulation data, the authors demonstrate that the model autonomously learns interatomic interaction strengths resembling classical electrostatics and identifies interaction cutoffs aligned with established physical models. The study also reports predictable neural scaling behavior and competitive accuracy on certain metrics compared to physics-informed architectures, suggesting that explicit priors may only be necessary when empirically justified.
arXiv:2607. 28304v1 Announce Type: new Abstract: Machine learning is transforming molecular sciences by accelerating property prediction, simulation, and the discovery of new molecules and materials.
arXiv:2510. 14217v2 Announce Type: replace Abstract: The spectral properties of feature embeddings offer critical insights into model generalization and representation quality.
arXiv:2606. 11382v1 Announce Type: new Abstract: Deep learning models facilitate the discovery of molecules with tailored properties among billions of candidate compounds.
arXiv:2606. 11508v1 Announce Type: new Abstract: Accurate prediction of absorption, distribution, metabolism, and excretion (ADME) properties is critical to drug discovery, but remains challenging because ADME endpoints are noisy, interdependent, and often data-limited.
arXiv:2602. 20573v3 Announce Type: replace Abstract: Molecules are often represented as SMILES strings, which can be readily converted to hand-crafted descriptors or fingerprints (FP) for molecular property prediction.
BOOM is a new benchmark for evaluating out‑of‑distribution (OOD) molecular property predictions in machine learning. It provides chemically‑informed tests across common property prediction tasks and assesses over 150 model‑task combinations. The study shows that current models, including chemical foundation models, struggle to generalize OOD, with the best model still exhibiting three times higher error than in‑distribution predictions.
arXiv:2606. 30961v1 Announce Type: cross Abstract: Advances in deep learning architectures and representations have enabled ML-driven chemical property prediction, but state-of-the-art (SOTA) models have remained largely confined to independent codebases and lack support for diverse chemical species.