A5034610400- Data Management and Algorithms
- Complex Network Analysis Techniques
- Advanced Clustering Algorithms Research
- Tensor decomposition and applications
- Complexity and Algorithms in Graphs
- Imbalanced Data Classification Techniques
- Functional Brain Connectivity Studies
- Advanced Neuroimaging Techniques and Applications
- Statistical and Computational Modeling
- Advanced Graph Neural Networks
- Parallel Computing and Optimization Techniques
- Anomaly Detection Techniques and Applications
- Face and Expression Recognition
- Crime Patterns and Interventions
- Advanced Database Systems and Queries
- Network Security and Intrusion Detection
- Machine Learning and Data Classification
- Neural Networks and Applications
- Image Retrieval and Classification Techniques
- Advanced Image and Video Retrieval Techniques
University of California, Davis
2009-2015
We pose the problem of network discovery which involves simplifying spatio-temporal data into cohesive regions (nodes) and relationships between those (edges). Such problems naturally exist in fMRI scans human subjects. These consist activations thousands voxels over time with aim to simplify them underlying cognitive being used. propose supervised semi-supervised variations this postulate a constrained tensor decomposition formulation corresponding alternating least squares solver that is...
Role discovery in graphs is an emerging area that allows analysis of complex intuitive way. In contrast to community discovery, which finds groups highly connected nodes, role nodes share similar topological structure the graph, and hence a common (or function) such as being broker or periphery node. However, existing work so far completely unsupervised, undesirable for number reasons. We provide alternating least squares framework convex constraints be placed on problem, can useful...
Hierarchical clustering is extensively used to organize high dimensional objects such as documents and images into a structure which can then be in multitude of ways. However, existing algorithms are limited their application since the time complexity agglomerative style much O(n2log n) where n number objects. Furthermore computation similarity between itself consuming given they dimension even optimized built functions found MATLAB take best part day handle collections just 10,000 on...
The area of constrained clustering has been actively pursued for the last decade. A more recent extension that will be focus this paper is hierarchical which allows building user-constrained dendrograms/trees. Like all forms clustering, previous work on uses simple constraints are typically implemented in a procedural language. However, there exists mature results and packages fields constraint satisfaction languages solvers field yet to explore. This marks first steps towards introducing...
Hierarchical clustering is typically implemented as a greedy heuristic algorithm with no explicit objective function. In this work we formalize hierarchical an integer linear programming (ILP) problem natural function and the dendrogram properties enforced constraints. Though exact solvers exists for ILP show that simple randomized (LP) relaxation can be used to provide approximate solutions faster. Formalizing also has benefit relaxing constraints produce novel variations such overlapping...
Role discovery in graphs is an emerging area that allows analysis of complex intuitive way. In contrast to other graph prob- lems such as community discovery, which finds groups highly connected nodes, the role problem nodes share similar topological structure. However, existing work so far has two severe limitations prevent its use some domains. Firstly, it completely unsupervised undesirable for a number reasons. Secondly, most limited single relational graph. We address both these lim-...