Abstract:The problem of learning the structure of a high dimensional graphical model from data has received considerable attention in recent years. In many applications such as sensor networks and proteomics it is often expensive to obtain samples from all the variables involved simultaneously. For instance, this might involve the synchronization of a large number of sensors or the tagging of a large number of proteins. To address this important issue, we initiate the study of a novel graphical model selection problem, where the goal is to optimize the total number of scalar samples obtained by allowing the collection of samples from only subsets of the variables. We propose a general paradigm for graphical model selection where feedback is used to guide the sampling to high degree vertices, while obtaining only few samples from the ones with the low degrees. We instantiate this framework with two specific active learning algorithms, one of which makes mild assumptions but is computationally expensive, while the other is more computationally efficient but requires stronger (nevertheless standard) assumptions. Whereas the sample complexity of passive algorithms is typically a function of the maximum degree of the graph, we show that the sample complexity of our algorithms is provable smaller and that it depends on a novel local complexity measure that is akin to the average degree of the graph. We finally demonstrate the efficacy of our framework via simulations.

Active Learning Algorithms for Graphical Model Selection

Active Learning for Undirected Graphical Model Selection

Lower Bounds on Active Learning for Graphical Model Selection

Active Model Selection for Graph-Based Semi-Supervised Learning

A Scalable Algorithm for Graph-Based Active Learning

S2: An Efficient Graph Based Active Learning Algorithm with Application to Nonparametric Classification

Structure Learning in Gaussian Graphical Models from Glauber Dynamics

Active Learning with Statistical Models

Efficient Learning of Discrete Graphical Models

Active Learning On Weighted Graphs Using Adaptive And Non-adaptive Approaches

Graph Active Learning at Subgraph Granularity

Robust Offline Active Learning on Graphs

Active Learning for Graphs with Noisy Structures

Active Learning on Attributed Graphs via Graph Cognizant Logistic Regression and Preemptive Query Generation

When Contrastive Learning Meets Active Learning: A Novel Graph Active Learning Paradigm with Self-Supervision

Adaptive Graph Active Learning with Mutual Information Via Policy Learning

Model Selection-inspired Coefficients Optimization for Polynomial-Kernel Graph Learning

A Structural-Clustering Based Active Learning for Graph Neural Networks

Active Learning for Hidden Attributes in Networks

MetaGL: Evaluation-Free Selection of Graph Learning Models via Meta-Learning

Deep Unsupervised Active Learning on Learnable Graphs