Hilbert-Geo creates the first unified formal language for solid geometry and a two-step parsing-then-reasoning method that reaches SOTA accuracy on solid geometry benchmarks.
super hub Mixed citations
Relational inductive biases, deep learning, and graph networks
Mixed citation behavior. Most common role is background (56%).
106
Pith papers citing it
Background
56% of classified citations
abstract
Artificial intelligence (AI) has undergone a renaissance recently, making major progress in key domains such as vision, language, control, and decision-making. This has been due, in part, to cheap data and cheap compute resources, which have fit the natural strengths of deep learning. However, many defining characteristics of human intelligence, which developed under much different pressures, remain out of reach for current approaches. In particular, generalizing beyond one's experiences--a hallmark of human intelligence from infancy--remains a formidable challenge for modern AI. The following is part position paper, part review, and part unification. We argue that combinatorial generalization must be a top priority for AI to achieve human-like abilities, and that structured representations and computations are key to realizing this objective. Just as biology uses nature and nurture cooperatively, we reject the false choice between "hand-engineering" and "end-to-end" learning, and instead advocate for an approach which benefits from their complementary strengths. We explore how using relational inductive biases within deep learning architectures can facilitate learning about entities, relations, and rules for composing them. We present a new building block for the AI toolkit with a strong relational inductive bias--the graph network--which generalizes and extends various approaches for neural networks that operate on graphs, and provides a straightforward interface for manipulating structured knowledge and producing structured behaviors. We discuss how graph networks can support relational reasoning and combinatorial generalization, laying the foundation for more sophisticated, interpretable, and flexible patterns of reasoning. As a companion to this paper, we have released an open-source software library for building graph networks, with demonstrations of how to use them in practice.
hub tools
citation-role summary
background 8
baseline 1
citation-polarity summary
claims ledger
- abstract Artificial intelligence (AI) has undergone a renaissance recently, making major progress in key domains such as vision, language, control, and decision-making. This has been due, in part, to cheap data and cheap compute resources, which have fit the natural strengths of deep learning. However, many defining characteristics of human intelligence, which developed under much different pressures, remain out of reach for current approaches. In particular, generalizing beyond one's experiences--a hallmark of human intelligence from infancy--remains a formidable challenge for modern AI. The followi
- background References [1] F. H. Clarke.Optimization and Nonsmooth Analysis. SIAM Classics in Applied Mathematics. SIAM, Philadelphia, 1990. [2] W. Ambrose and I. M. Singer.A theorem on holonomy.Transactions of the American Mathematical Society, 75(3):428-443, 1953. [3] J. L. Ba, J. R. Kiros, and G. E. Hinton.Layer normalization.stat, 1050. Jg., S. 21., 2016. [4] P. W. Battaglia et al.Relational inductive biases, deep learning, and graph networks.arXiv preprint arXiv:1806.01261, 2018. [5] M. M. Bronstein, J
- baseline customer-transaction bipartite graph. The architecture consists of a GNN encoder trained on a user profile feature set; and a feed- forward decoder built for the task of anomaly detection. FraudGT [19] is the most recent graph-based transformer method. In their pa- per, the authors compare their model performance with a variety of GNN-based models [2, 5, 8, 10, 13, 14, 24, 31, 35, 36]. They prove the superiority of their model over all these methods. FraudGT [19] and MultiGNN [10] are also two o
- background yields state-of-the-art accuracy among Vision SSMs on ImageNet-1K, COCO, and ADE20K at comparable parameter and FLOP budgets. Because GraphScan sits on the input side, it is orthogonal to recurrence-side advances [46, 68, 6, 27]; composing both directions is a natural next step. To the question posed in the title: yes-graphs do help Vision SSMs see better. 9 References [1] Peter W Battaglia, Jessica B Hamrick, Victor Bapst, Alvaro Sanchez-Gonzalez, Vinicius Zambaldi, Mateusz Malinowski, Andrea T
- background RQ1 - Graph as the prior interface (built on P1). Existing approaches inject domain priors into deep time-series models implicitly, through architectural choices: decomposition blocks [4, 5], sparse attention patterns [6], or task-specific spatio-temporal networks [20]. Each new type of prior demands a new architectural trick, with no shared interface [21, 22]. Primitive P1 turns the factor graph itself into the prior interface, raising the question:can well-known symbolic time-series priors-e.g
- background for capturing evolving coordination patterns over time, while AGP [37] introduces action-level graph priors that explicitly model dependencies between agent-action pairs to improve fine-grained coordination. Beyond coordination graphs, graph neural networks have been applied more broadly in MARL to model agent interactions [38, 39], encode relational inductive biases [40], and support scalable multi-agent communica- tion [41]. Graph attention networks [42] and graph trans- formers [12] have been
- background statistical alignment between source and target domains, a process often reliant on a large source dataset [58]; thus, its success in depression detection [34] may depend heavily on the transferred knowledge. In contrast, the success of the CDMA framework stems not from inheriting knowledge but from its inherent architectural design. This provides a crucial inductive bias [59] that allows it to naturally discover the shared pattern of depression even within a typologically distinct language. Thi
authors
co-cited works
representative citing papers
CPF-GCD enforces low-rank compositional structure on vision backbone features via spatial primitive fields so that novel categories emerge as new activation patterns over a shared vocabulary of reusable visual primitives.
GraphNPE recovers a significantly lower central density for Boötes I consistent with a core while Draco remains marginally cuspy, and demonstrates that higher-order velocity moments reduce bias in dynamical modeling.
ConSolv is a solvent-conditional attention-based MLP trained on experimental and ab initio solvation free energies that generalizes across 66 organic solvents and matches some experimental NMR data.
DS-HGNN achieves lower RMSE for stress and displacement prediction on stiffened panels than six benchmark GNN models and matches top accuracy with 19-38% fewer training samples.
ReSYNC learns recovery skills via RL then discovers and refines relational predicates to enable abstract planning that generalizes failure avoidance to unseen long-horizon tasks, outperforming baselines by over 50% in simulation and transferring to real robots.
Attention flows as dynamic contextualised representations recover analogy relationships among research outputs that count-based and sequence baselines fail to recover.
Face-Feature Tuning is a label-free logit remapping method that reduces FPR/TPR gaps across groups in deepfake detection while preserving overall accuracy.
Introduces graph-to-image prediction of per-node dynamic stability landscapes in oscillator networks from topology, releases two 10k-graph datasets, and shows GNN-CNN models achieve good accuracy with cross-size generalization.
Introduces transitive inference with exceptions task and analytically shows kernel ridge regression balances relational generalization and memorization depending on representational geometry, with validation in finetuned language models.
GraphScan replaces geometric or coordinate-based scanning in Vision SSMs with learned local semantic graph routing, yielding SOTA results among such models on classification and segmentation tasks.
Graph neural networks can approximate full 3D non-LTE Ca II populations in solar models with correlations above 0.99 and extreme computational efficiency.
NEO is a probabilistic neural model that induces compositional programs as a learned Language of Thought from non-textual observations and executes them via a shared transition model to enable explanation-driven generalization.
The paper unifies emerging graph-based world models under a new paradigm and proposes a taxonomy organized by spatial, physical, and logical relational inductive biases.
PiGGO integrates a learned graph neural ODE as the continuous-time dynamics model within an extended Kalman filter to enable online virtual sensing and uncertainty-aware state estimation for nonlinear dynamic systems with unknown model form and sparse sensing.
Scale-autoregressive modeling (SAR) samples fluid flow distributions hierarchically from coarse to fine resolutions on meshes, achieving lower distributional error and 2-7x faster runtime than diffusion or flow-matching baselines.
A self-supervised multimodal alignment step plus equivariant GNN-based MARL yields over twofold sensing accuracy and 50% performance gains in decentralized V2I rate maximization.
Smoothness assumptions on graphical model kernels produce Wasserstein estimation rates determined by local graph structure rather than ambient dimension.
Causal Process Models reframe dynamic causal graph discovery as multi-agent reinforcement learning to build sparse time-varying graphs only at active interactions, outperforming dense baselines on physical prediction.
In-weights learning induces linear embeddings enabling transitive inference in transformers, whereas in-context learning defaults to match-and-copy unless pre-trained on linear tasks or prompted with linear mental maps.
Unsupervised GNN model learns local updates for approximate MaxIS on dynamic graphs, achieving competitive ratios on 200-1000 node instances and 1.00-1.18x larger solutions than other unsupervised models when generalizing to 100x larger graphs.
Temporal Graph Networks combine memory modules and graph operators to learn on dynamic graphs as timed event sequences, outperforming prior methods on transductive and inductive tasks while unifying earlier models as special cases.
Graph Kernel Networks learn PDE solution operators that generalize across discretization methods and grid resolutions using graph-based kernel integration.
BERT stores relational knowledge extractable via cloze queries without fine-tuning and matches supervised baselines on open-domain QA tasks.
citing papers explorer
-
Representing Research Attention as Contextually Structured Flows
Attention flows as dynamic contextualised representations recover analogy relationships among research outputs that count-based and sequence baselines fail to recover.
-
Language Models as Knowledge Bases?
BERT stores relational knowledge extractable via cloze queries without fine-tuning and matches supervised baselines on open-domain QA tasks.
-
Graph-Based Alternatives to LLMs for Human Simulation
GEMS formulates close-ended human-behavior simulation as link prediction on a heterogeneous graph and matches or exceeds LLM performance with three orders of magnitude fewer parameters across three datasets and three evaluation settings.
-
Grammatically-Guided Sparse Attention for Efficient and Interpretable Transformers
Grammatically-Guided Sparse Attention uses POS tags to generate hard or soft masks that constrain self-attention, achieving 0.8200 and 0.8165 accuracy on SST-2 versus 0.8200 for full attention in a DistilBERT-like model.