SenCos-GEM: SENet-Calibrated and Law-of-Cosines-Constrained Geometry-Enhanced Molecular Representation for Property Prediction

arXiv cs.LG Papers
molecular-representation property-prediction 3d-gnn self-supervised-learning geometry-enhanced senet law-of-cosines

Summary

SenCos-GEM introduces a physics-guided molecular representation learning framework that uses Squeeze-and-Excitation modules and a law-of-cosines constraint to improve 3D geometric understanding, achieving state-of-the-art results on MoleculeNet regression benchmarks.

arXiv:2607.20551v1 Announce Type: new Abstract: Effective molecular representation learning is crucial for accurate molecular property prediction. Recently, numerous self-supervised learning (SSL) approaches leveraging 3D GNNs have been developed to capture comprehensive 3D structural information for drug discovery. However, existing methods lack explicit physical constraints and are highly susceptible to geometric noise induced by coarse empirical force fields during large-scale pre-training.Furthermore, they overlook dynamic feature modulation during downstream adaptation, often resulting in catastrophic forgetting and negative transfer. To address these limitations, we introduce SenCos-GEM, a novel explicitly decoupled geometry-enhanced molecular representation learning framework that incorporates SENet-calibrated and law-of-cosines-constrained enhancements. SenCos-GEM employs a physics-guided geometric consistency loss based on the law of cosines to derive high-fidelity and mathematically invariant 3D spatial priors. In addition, lightweight Squeeze-and-Excitation (SE) modules are integrated into the backbone as task-specific adapters, while a dual-modulation prediction head combines Feature-wise Linear Modulation (FiLM) and SENet mechanisms to enable dynamic feature recalibration. SenCos-GEM demonstrates highly competitive performance across diverse classification and regression tasks on MoleculeNet benchmark, establishing new state-of-the-art results specifically on 3D conformation-sensitive regression tasks, such as FreeSolv, Lipophilicity, and QM9, achieving relative error reductions of 12.9% (RMSE), 5.3% (RMSE), and 8.2% (MAE), respectively. Moreover, our model exhibits superior capability in distinguishing stereoisomers and discriminating conformational perturbations, underscoring its robust spatial modeling performance. Collectively, SenCos-GEM represents a significant breakthrough in accurate molecular property prediction.
Original Article
View Cached Full Text

Cached at: 07/24/26, 05:12 AM 

# SenCos-GEM: SENet-Calibrated and Law-of-Cosines-Constrained Geometry-Enhanced Molecular Representation for Property Prediction
Source: [https://arxiv.org/abs/2607.20551](https://arxiv.org/abs/2607.20551)
[View PDF](https://arxiv.org/pdf/2607.20551)

> Abstract:Effective molecular representation learning is crucial for accurate molecular property prediction\. Recently, numerous self\-supervised learning \(SSL\) approaches leveraging 3D GNNs have been developed to capture comprehensive 3D structural information for drug discovery\. However, existing methods lack explicit physical constraints and are highly susceptible to geometric noise induced by coarse empirical force fields during large\-scale[this http URL](http://pre-training.furthermore/), they overlook dynamic feature modulation during downstream adaptation, often resulting in catastrophic forgetting and negative transfer\. To address these limitations, we introduce SenCos\-GEM, a novel explicitly decoupled geometry\-enhanced molecular representation learning framework that incorporates SENet\-calibrated and law\-of\-cosines\-constrained enhancements\. SenCos\-GEM employs a physics\-guided geometric consistency loss based on the law of cosines to derive high\-fidelity and mathematically invariant 3D spatial priors\. In addition, lightweight Squeeze\-and\-Excitation \(SE\) modules are integrated into the backbone as task\-specific adapters, while a dual\-modulation prediction head combines Feature\-wise Linear Modulation \(FiLM\) and SENet mechanisms to enable dynamic feature recalibration\. SenCos\-GEM demonstrates highly competitive performance across diverse classification and regression tasks on MoleculeNet benchmark, establishing new state\-of\-the\-art results specifically on 3D conformation\-sensitive regression tasks, such as FreeSolv, Lipophilicity, and QM9, achieving relative error reductions of 12\.9% \(RMSE\), 5\.3% \(RMSE\), and 8\.2% \(MAE\), respectively\. Moreover, our model exhibits superior capability in distinguishing stereoisomers and discriminating conformational perturbations, underscoring its robust spatial modeling performance\. Collectively, SenCos\-GEM represents a significant breakthrough in accurate molecular property prediction\.

## Submission history

From: Tianming Han \[[view email](https://arxiv.org/show-email/c9fcf230/2607.20551)\] **\[v1\]**Wed, 15 Jul 2026 00:23:51 UTC \(1,454 KB\)

Similar Articles

GLACIER: A Multimodal Student-Teacher Foundation Model for Molecular Property Prediction

arXiv cs.LG

This paper introduces GLACIER, a multimodal student-teacher foundation model that integrates molecular graphs, SMILES strings, and physicochemical descriptors to predict molecular properties efficiently. It leverages Finsler geometry-aware fusion and knowledge distillation from larger teacher models (MiniMol, MolFormer) to achieve high performance with a lightweight architecture.

GEM: Geometric Entropy Mixing for Optimal LLM Data Curation

arXiv cs.LG

GEM reformulates LLM data curation as a variational problem on the hypersphere, using geometric entropy mixing and a minorize-maximize algorithm to discover balanced semantic clusters, achieving state-of-the-art improvements in data mixing strategies by up to 1.2% average downstream accuracy.