Metric Distribution to Vector: Constructing Data Representation via Broad-Scale Discrepancies

TL;DR

This paper introduces MetricDistribution2vec, a novel graph embedding method that captures dataset-wide metric distribution discrepancies to improve classification accuracy, especially in few-shot scenarios.

Contribution

The paper proposes a new embedding strategy that encodes distribution characteristics of data discrepancies, enhancing graph classification performance over existing methods.

Findings

Significant accuracy improvements over baseline methods on real-world datasets.

Effective in few-shot classification with limited training samples.

Lightweight models achieve competitive results.

Abstract

Graph embedding provides a feasible methodology to conduct pattern classification for graph-structured data by mapping each data into the vectorial space. Various pioneering works are essentially coding method that concentrates on a vectorial representation about the inner properties of a graph in terms of the topological constitution, node attributions, link relations, etc. However, the classification for each targeted data is a qualitative issue based on understanding the overall discrepancies within the dataset scale. From the statistical point of view, these discrepancies manifest a metric distribution over the dataset scale if the distance metric is adopted to measure the pairwise similarity or dissimilarity. Therefore, we present a novel embedding strategy named $\mathbf{MetricDistribution2vec}$ to extract such distribution characteristics into the vectorial representation for each data. We demonstrate the application and effectiveness of our representation method in the supervised prediction tasks on extensive real-world structural graph datasets. The results have gained some unexpected increases compared with a surge of baselines on all the datasets, even if we take the lightweight models as classifiers. Moreover, the proposed methods also conducted experiments in Few-Shot classification scenarios, and the results still show attractive discrimination in rare training samples based inference.