A5021862004- Complex Network Analysis Techniques
- Advanced Graph Neural Networks
- Functional Brain Connectivity Studies
- Graph Theory and Algorithms
- Bayesian Modeling and Causal Inference
- Data Management and Algorithms
- Bioinformatics and Genomic Networks
- Opinion Dynamics and Social Influence
- Face and Expression Recognition
- Gene Regulatory Network Analysis
- Advanced Neuroimaging Techniques and Applications
- Neural dynamics and brain function
- Topological and Geometric Data Analysis
- Neurobiology and Insect Physiology Research
- Morphological variations and asymmetry
- Data-Driven Disease Surveillance
- Statistical Methods and Inference
- Natural Language Processing Techniques
- Neural Networks and Applications
- Insect and Arachnid Ecology and Behavior
- Data Visualization and Analytics
- Advanced Clustering Algorithms Research
- Anomaly Detection Techniques and Applications
- Random Matrices and Applications
- Bayesian Methods and Mixture Models
Johns Hopkins University
2016-2025
Mathematical Institute of the Slovak Academy of Sciences
2023
Redmond Fire Department
2023
University of Delaware
2023
University of Baltimore
2022
North Carolina State University
2020
George Washington University
2011
A single nervous system can generate many distinct motor patterns. Identifying which neurons and circuits control behaviors has been a laborious piecemeal process, usually for one observer-defined behavior at time. We present fundamentally different approach to neuron-behavior mapping. optogenetically activated 1054 identified neuron lines in Drosophila larvae tracked the behavioral responses from 37,780 animals. Application of multiscale unsupervised structure learning methods data enabled...
Brains contain networks of interconnected neurons and so knowing the network architecture is essential for understanding brain function. We therefore mapped synaptic-resolution connectome an entire insect ( Drosophila larva) with rich behavior, including learning, value computation, action selection, comprising 3016 548,000 synapses. characterized neuron types, hubs, feedforward feedback pathways, as well cross-hemisphere brain-nerve cord interactions. found pervasive multisensory...
In disciplines as diverse social network analysis and neuroscience, many large graphs are believed to be composed of loosely connected smaller graph primitives, whose structure is more amenable We propose a robust, scalable, integrated methodology for community detection comparison in graphs. our procedure, we first embed into an appropriate Euclidean space obtain low-dimensional representation, then cluster the vertices communities. next employ nonparametric inference techniques identify...
The ability to detect change-points in a dynamic network or time series of graphs is an increasingly important task many applications the emerging discipline graph signal processing. This paper formulates change-point detection as hypothesis testing problem terms generative latent position model, focusing on special case Stochastic Block Model series. We analyze two classes scan statistics, based distinct underlying locality statistics presented literature. Our main contribution derivation...
Two-sample hypothesis testing for random graphs arises naturally in neuroscience, social networks, and machine learning. In this article, we consider a semiparametric problem of two-sample class latent position graphs. We formulate notion consistency context propose valid test the that two finite-dimensional dot product on common vertex set have same generating positions or are scaled diagonal transformations one another. Our statistic is function spectral decomposition adjacency matrix each...
The random dot product graph (RDPG) is an independent-edge that analytically tractable and, simultaneously, either encompasses or can successfully approximate a wide range of graphs, from relatively simple stochastic block models to complex latent position graphs. In this survey paper, we describe comprehensive paradigm for statistical inference on centered spectral embeddings adjacency and Laplacian matrices. We examine the analogues, in inference, several canonical tenets classical...
Significance Spectral graph clustering—clustering the vertices of a based on their spectral embedding—is significant current interest, finding applications throughout sciences. But as with clustering in general, what particular methodology identifies “clusters” is defined (explicitly, or, more often, implicitly) by algorithm itself. We provide clear and concise demonstration “two-truths” phenomenon for which first step—spectral either Laplacian embedding, wherein one decomposes normalized...
Given a time series of graphs G(t) = (V, E(t)), t 1, 2, ..., where the fixed vertex set V represents "actors" and an edge between u v at (uv \in E(t)) existence communications event actors during tth period, we wish to detect anomalies and/or change points. We consider collection graph features, or invariants, demonstrate that adaptive fusion provides superior inferential efficacy compared naive equal weighting for certain class anomaly detection problems. Simulation results using latent...
Genome-wide association studies have demonstrated significant links between human brain structure and common DNA variants. Similar with rodents been challenging because of smaller volumes. Using high field MRI (9.4 T) compressed sensing, we achieved microscopic resolution sufficiently throughput for rodent population studies. We generated whole structural diffusion connectomes four diverse isogenic lines mice (C57BL/6J, DBA/2J, CAST/EiJ, BTBR) at spatial 20,000 times higher than connectomes....
In this paper, we introduce the concept of principal communities and propose a graph encoder embedding method that concurrently detects these achieves vertex embedding. Given adjacency matrix with labels, computes sample community score for each community, ranking them to measure importance estimate set communities. The then produces by retaining only dimensions corresponding Theoretically, define population version based on random Bernoulli distribution. We prove preserves conditional...
Abstract This article describes a large multi‐institutional analysis of the shape and structure human hippocampus in aging brain as measured via MRI. The study was conducted on population 101 subjects including nondemented control ( n = 57) clinically diagnosed with Alzheimer's Disease (AD, 38) or semantic dementia 6) imaging data collected at Washington University St. Louis, hippocampal annotated Massachusetts General Hospital, anatomical shapes embedded into metric space using deformation...
Performing statistical analyses on collections of graphs is import to many disciplines, but principled, scalable methods for multi-sample graph inference are few. Here we describe an "omnibus" embedding in which multiple the same vertex set jointly embedded into a single space with distinct representation each graph. We prove central limit theorem this and demonstrate how it streamlines comparison, obviating need pairwise subspace alignments. The omnibus achieves near-optimal accuracy when...
Analyzing large-scale time-series network data, such as social media and email communications, poses a significant challenge in understanding dynamics, detecting anomalies, predicting trends. In particular, the scalability of graph analysis is critical hurdle impeding progress downstream inference. To address this challenge, we introduce temporal encoder embedding method. This approach leverages ground-truth or estimated vertex labels, enabling an efficient data processing billions edges...
The problem of finding the vertex correspondence between two noisy graphs with different number vertices where smaller graph is still large has many applications in social networks, neuroscience, and computer vision. We propose a solution to this via matching matched filter: centering padding adjacency matrix applying methods align it larger network. schemes can be incorporated into any algorithm that matches using matrices. Under statistical model for correlated pairs graphs, which yields...
Our problem of interest is to cluster vertices a graph by identifying underlying community structure. Among various vertex clustering approaches, spectral one the most popular methods because it easy implement while often outperforming more traditional algorithms. However, there are two inherent model selection problems in clustering, namely estimating both embedding dimension and number clusters. This article attempts address issue establishing novel framework specifically for on graphs...
We define a latent structure random graph as dot product (RDPG) in which the position distribution incorporates both probabilistic and geometric constraints, delineated by family of underlying distributions on some fixed Euclidean space, structural support submanifold from are drawn positions for graph. For one-dimensional model with known support, we extend existing results consistency spectral estimates RDPGs to demonstrate that parameters can be efficiently estimated. describe how...
In this paper, we introduce a novel and computationally efficient method for vertex embedding, community detection, size determination. Our approach leverages normalized one-hot graph encoder rank-based cluster measure. Through extensive simulations, demonstrate the excellent numerical performance of our proposed ensemble algorithm.
We present a novel approximate graph matching algorithm that incorporates seeded data into the paradigm. Our Joint Optimization of Fidelity and Commensurability (JOFC) embeds two graphs common Euclidean space where inference task can be performed. Through real simulated examples, we demonstrate versatility our in with various characteristics--weightedness, directedness, loopiness, many-to-one many-to-many matchings, soft seedings.
Hypothesis testing on time series of attributed graphs has applications in diverse areas, e.g., social network analysis (wherein vertices represent individual actors or organizations), connectome inference are neurons brain regions) and text processing authors documents). We consider the problem anomaly/change point detection given latent process model for with categorical attributes edges presented [N. H. Lee C. E. Priebe, “A random graphs,” Statist. Inference Stoch. Process., vol. 14, pp....