Research
On-device research index

arXiv research

A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.

168,657 papers · 148 categories

Trend · papers per month

100201301401 · Jun 202019922001200920172026
48 results for graph topological gap

This paper refines understanding of decentralized learning by considering graph topology.

problem Current theory fails to predict performance in decentralized learning settings.
method Quantifies how graph topology influences convergence in decentralized learning.
result Graph topology significantly impacts convergence in decentralized learning, contrary to spectral gap theory.

A new topology improves decentralized learning efficiency and accuracy.

problem Finding efficient decentralized learning topologies with fast consensus and low maximum degree.
method Proposed the Base-(k+1)(k + 1) Graph topology for decentralized learning.
result The Base-(k+1)(k + 1) Graph enables faster convergence and better communication efficiency than the exponential graph.

Persistent homology enhances graph classification by capturing long-range graph properties.

problem Lack of formal assessment of persistent homology in graph learning.
method Brief introduction and theoretical discussion of persistent homology in graph context, followed by empirical analysis.
result Persistent homology improves graph classification, especially for data with prominent topological structures.

The paper classifies CMC free boundary hypersurfaces in rotational domains.

problem Existence and uniqueness of free boundary constant mean curvature hypersurfaces in rotational domains.
method Classification and construction of CMC free boundary hypersurfaces under specific conditions.
result Classification of CMC free boundary hypersurfaces as topological disks or annuli.

Graph pruning improves neural network performance by addressing squashing and smoothing issues.

problem Over-squashing and over-smoothing in Graph Neural Networks.
method Proposes edge deletions to simultaneously address over-squashing and over-smoothing, optimizing spectral gap.
result Edge deletions improve generalization and distinguishability of nodes of different classes.

We study the spectral gap of the Erdős--Rényi random graph through the connectivity threshold. In particular, we show that for any fixed δ>0δ> 0 if p(1/2+δ)lognn,p \ge \frac{(1/2 + δ) \log n}{n}, then the normalized graph Laplacian of an Erdős--Rényi graph has all of its nonzero eigenvalues tightly concentrated around 11. We est…

2012-01-02abs ↗pdf ↗

AdaCGP learns dynamic graph topology from time series data, improving over existing methods.

problem Learning dynamic graph topology from time-varying signals, especially in real-time applications.
method AdaCGP is a sparsity-aware adaptive algorithm that recursively estimates the Graph Shift Operator (GSO) through variable splitting.
result AdaCGP outperforms state-of-the-art methods in GSO estimation, achieving improvements exceeding 83%.

Graph Laplacians and machine learning predict properties of finite graphs.

problem Understanding properties of finite graphs using spectral and topological methods.
method Combining graph Laplacians, spectral inequalities, machine learning, and topological data analysis.
result Neural networks can accurately predict graph properties like Ricci-flatness and spectral gaps.

FoSR adds edges to graphs to prevent oversquashing and oversmoothing in GNNs.

problem Oversquashing and oversmoothing in graph neural networks (GNNs).
method First-order spectral rewiring to add edges based on spectral expansion, combined with a relational architecture.
result Our algorithm outperforms existing graph rewiring methods in graph classification tasks.

We propose Sparse Neural Network architectures that are based on random or structured bipartite graph topologies. Sparse architectures provide compression of the models learned and speed-ups of computations, they can also surpass their unstructured or fully connected counterparts. As we show, even more compact topologi…

2017-06-18abs ↗pdf ↗

Solved Cheeger inequalities for simplicial complexes, combining topological and graph theoretic methods.

problem Extend Cheeger inequalities to simplicial complexes and their higher order Laplacians.
method Combining constructions from simplicial topology, signed graphs, Gromov filling radii, and interpolating between 1-Laplacians and 2-Laplacians.
result Developed a general theory for p-Laplacians on simplicial complexes and proved Cheeger-type inequalities.

Paper proposes G-CRD to improve GNNs by preserving global graph topology.

problem Improving lightweight GNNs for robust performance on large-scale real-world graphs.
method Introduces Graph Contrastive Representation Distillation (G-CRD) using contrastive learning.
result G-CRD consistently boosts GNN performance and robustness, outperforming existing methods.

Study on stability of GCNNs under graph perturbations.

problem Limited theoretical understanding of GCNN stability.
method Proposes a probabilistic framework to analyze GCNN stability under various graph perturbations.
result Demonstrates the importance of data distribution in stability analysis.

Graph partitioning is the problem of dividing the nodes of a graph into balanced partitions while minimizing the edge cut across the partitions. Due to its combinatorial nature, many approximate solutions have been developed, including variants of multi-level methods and spectral clustering. We propose GAP, a Generaliz…

2019-03-02abs ↗pdf ↗

The paper derives upper bounds on the MLE error for BTL model under general graphs.

problem Estimating the MLE of BTL model parameters with \ell_{\infty}-loss under general graphs.
method Novel upper bounds on \ell_{\infty} estimation error dependent on algebraic connectivity and graph topology.
result Upper bounds on \ell_{\infty} error are sharp and match minimax lower bounds under certain graph topologies.

GCNs improve regression tasks by aggregating neighbor signals.

problem GCNs' statistical properties in regression tasks are poorly understood.
method Examined two GCN convolutions and their impact on learning error.
result GCNs have a bias-variance trade-off that depends on neighborhood size and topology.

This paper develops a discrete theory of real Riemann surfaces using quad-graphs and linear discretization.

problem Constructing a discrete theory of real Riemann surfaces.
method Using quad-graphs and linear discretization of Cauchy-Riemann equations, constructing a symplectic homology basis.
result The discrete period matrix has the same canonical decomposition as in the smooth setting.

SM-netFusion estimates brain network atlas by considering multiple topological measures.

problem Limited BNA estimation methods that overlook topological measures and lack discriminative power.
method Supervised multi-topology network cross-diffusion framework using degree, closeness, and eigenvector centrality measures.
result SM-netFusion produces more centered and representative templates, and improves classification accuracy.

The paper constructs noncompact hyperbolic surfaces with uniform spectral gaps using random graph models.

problem Building noncompact hyperbolic surfaces with uniform spectral gaps.
method Introduced a random graph model Fχ,n\mathcal{F}_{χ,n} to construct expanding families of graphs, then applied these families to create hyperbolic surfaces.
result Explicitly constructed an expanding family of graphs in the critical regime, leading to a sequence of complete, noncompact hyperbolic surfaces with uniformly positive spectral gaps.

Improved phylogenetic inference using VBPI-Mixtures for tree topology and branch length.

problem Multimodality of tree-topology posterior distributions in phylogenetic inference.
method VBPI-Mixtures algorithm that uses mixture learning within the BBVI framework.
result VBPI-Mixtures captures tree-topology distributions better than VBPI.

Statistical-computational gap found in aligning multiple Gaussian graphs.

problem Aligning multiple Gaussian graphs with unknown signals.
method Generalized informational threshold and computational barrier analysis.
result Existence of a statistical-computational gap in multiple Gaussian graph alignment.

The "least absolute shrinkage and selection operator" (Lasso) method has been adapted recently for networkstructured datasets. In particular, this network Lasso method allows to learn graph signals from a small number of noisy signal samples by using the total variation of a graph signal for regularization. While effic…

2017-04-07abs ↗pdf ↗

We prove that the spectral gap of a finite planar graph XX is bounded by $λ_1(X)\le C(\frac{\log(\diam X)}{\diam X})^2$ where CC depends only on the degree of XX. We then give a sequence of such graphs showing the the above estimate cannot be improved. This yields a negative answer to a question of Benjamini and Cur…

2012-04-19abs ↗pdf ↗

This paper is first-line research expanding GANs into graph topology analysis. By leveraging the hierarchical connectivity structure of a graph, we have demonstrated that generative adversarial networks (GANs) can successfully capture topological features of any arbitrary graph, and rank edge sets by different stages a…

2017-07-19abs ↗pdf ↗

Study on stable commutator length in RAAGs and Coxeter groups, proving spectral gaps and hardness results.

problem Understanding stable commutator length in right-angled Artin and Coxeter groups.
method Established spectral gaps, determined sizes up to constants, and related to graph properties.
result Found that stable commutator length can be arbitrarily close to zero in some groups, contrasting uniform gaps.

Unified taxonomy for graph representation learning.

problem Lack of unified understanding and integration of graph representation learning methods.
method Proposes a Graph Encoder Decoder Model (GRAPHEDM) to unify graph neural networks, network embedding, and graph regularization.
result Unified taxonomy and Graph Encoder Decoder Model (GRAPHEDM) for graph representation learning.

GNNs may be limited by graph topology, affecting their learning outcomes.

problem Understanding how graph topology influences GNN behavior and performance.
method Investigating the interaction between local topological features and GNN message-passing schemes.
result Locally similar neighborhoods can lead to consistent node representations, affecting GNN performance.

Novel TRI-GNN framework improves graph classification robustness.

problem Graph neural networks suffer from over-smoothing and vulnerability to graph perturbations.
method Integrates higher-order graph information via persistent homology and local graph structure learning.
result TRI-GNN outperforms state-of-the-art baselines on node classification tasks.

Detects graph topology changes from noisy signals using prior spectral information.

problem Detecting changes in graph topology from graph signals.
method Leverages graph filtering and subspace detection to distill problem into a CUSUM-based algorithm.
result Demonstrates the effectiveness of incorporating prior spectral signatures for change-point detection.

We consider efficiency in the implementation of deep neural networks. Hardware accelerators are gaining interest as machine learning becomes one of the drivers of high-performance computing. In these accelerators, the directed graph describing a neural network can be implemented as a directed graph describing a Boolean…

2019-05-07abs ↗pdf ↗