Explain Graph Neural Networks to Understand Weighted Graph Features in Node Classification

Li, Xiaoxiao; Saúde, João

doi:10.48550/arxiv.2002.00514

Cited by 2 publications

(3 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Interpreting graph convolutional networks Finally, a paper of interest is [37]. They apply graph neural networks with edge weights to networks of financial transactions, making node classification experiments on the Bitcoin network.…”

Section: Graph Modellingmentioning

confidence: 99%

Competition analysis on the over-the-counter credit default swap market

Abraham

2020

Preprint

View full text Add to dashboard Cite

We study two questions related to competition on the OTC CDS market using data collected as part of the EMIR regulation.First, we wanted to study the competition between central counterparties through collateral requirements. We present models that successfully estimate the initial margin requirements. However, our estimations are not precise enough to use them as input to a predictive model for CCP choice by counterparties in the OTC market.Second, we model counterpart choice on the interdealer market using a novel semi-supervised predictive task. We present our methodology as part of the literature on model interpretability before arguing for the use of conditional entropy as the metric of interest to derive knowledge from data through a model-agnostic approach. In particular, we justify the use of deep neural networks to measure conditional entropy on real-world datasets. We created the Razor entropy using the framework of algorithmic information theory and derived an explicit formula that is identical to our semi-supervised training objective. Finally, we borrowed concepts from game theory to define top-k Shapley values. This novel method of payoff distribution satisfies most of the properties of Shapley values, and is of particular interest when the value function is monotone submodular. Unlike (classical) Shapley values, top-k Shapley values can be computed in quadratic time of the number of features instead of exponential. We implemented our methodology and report the results on our particular task of counterpart choice.Finally, we present an improvement to the node2vec algorithm that we intended to use to study the intermediation in the OTC market. We show that the neighbor sampling used in the generation of biased walks can be performed in logarithmic time with a quasilinear time (and space) pre-computation, unlike the current implementations that do not scale well.

show abstract

Section: Graph Modellingmentioning

confidence: 99%

Competition analysis on the over-the-counter credit default swap market

Abraham

2020

Preprint

View full text Add to dashboard Cite

show abstract

“…In fact, there have been several attempts [3][4][5][6][7][8][9] in the literature to open the black box of the neural message passing. Paper [3] attributed the success of GNN to Laplacian smoothing, which makes the features of vertices in the same cluster similar and thus easy for clustering.…”

Section: Introductionmentioning

confidence: 99%

“…The remaining interpretations [6][7][8][9] focus on graph structures or features, and try to identify the informative components and important node features which have a crucial role in GNN's prediction. However, when M or X is random, if M k X still contributes to the community detection, their interpretations may need some patches.…”

Section: Introductionmentioning

confidence: 99%

Understanding the Message Passing in Graph Neural Networks via Power Iteration Clustering

Li,

Cheng

2020

Preprint

View full text Add to dashboard Cite

The mechanism of message passing in graph neural networks(GNNs) is still mysterious for the literature. No one, to our knowledge, has given another possible theoretical origin for GNNs apart from convolutional neural networks. Somewhat to our surprise, the message passing can be best understood in terms of the power iteration. By removing activation functions and layer weights of GNNs, we propose power iteration clustering (SPIC) models which are naturally interpretable and scalable. The experiment shows our models extend the existing GNNs and enhance its capability of processing random featured networks. Moreover, we demonstrate the redundancy of some state-of-the-art GNNs in designing and define a lower limit for model evaluation by randomly initializing the aggregator of message passing. All the findings in this paper push the boundaries of our understanding of neural networks.

show abstract

Explain Graph Neural Networks to Understand Weighted Graph Features in Node Classification

Cited by 2 publications

References 10 publications

Competition analysis on the over-the-counter credit default swap market

Competition analysis on the over-the-counter credit default swap market

Understanding the Message Passing in Graph Neural Networks via Power Iteration Clustering

Contact Info

Product

Resources

About