Experimental analyses on 2-hop-based and 3-hop-based link prediction algorithms

Zhou, Tao; Lee, Yan-Li; Wang, Guannan

doi:10.1016/j.physa.2020.125532

Cited by 48 publications

(33 citation statements)

References 55 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…(2020) for more CH indices according to the core idea). Extensive empirical analyses ( Muscoloni et al., 2020 ; Zhou et al., 2021 ) indicated that the introduction of local community paradigm and Hebbian learning rule could considerably improve the performance of routine local similarity indices.…”

Section: Local Similarity Indicesmentioning

confidence: 99%

“…We have implemented extensive experiments on 137 real networks ( Zhou et al., 2021 ), suggesting that (i) 3-hop-based indices outperform 2-hop-based indices subject to AUC, while 3-hop-based and 2-hop-based indices are competitive on precision; (ii) CH indices perform the best among all considered candidates; and (iii) 3-hop-based indices are more suitable for disassortative networks with lower densities and lower average clustering coefficient. Furthermore, we have showed that a hybrid of 2-hop-based and 3-hop-based indices via collaborative filtering techniques can result in overall better performance ( Lee and Zhou, 2021 ).…”

Section: Local Similarity Indicesmentioning

confidence: 99%

See 1 more Smart Citation

Progresses and challenges in link prediction

Zhou

2021

iScience

Self Cite

View full text Add to dashboard Cite

Summary Link prediction is a paradigmatic problem in network science, which aims at estimating the existence likelihoods of nonobserved links, based on known topology. After a brief introduction of the standard problem and evaluation metrics of link prediction, this review will summarize representative progresses about local similarity indices, link predictability, network embedding, matrix completion, ensemble learning, and some others, mainly extracted from related publications in the last decade. Finally, this review will outline some long-standing challenges for future studies.

show abstract

Section: Local Similarity Indicesmentioning

confidence: 99%

Section: Local Similarity Indicesmentioning

confidence: 99%

Progresses and challenges in link prediction

Zhou

2021

iScience

Self Cite

View full text Add to dashboard Cite

show abstract

“…It might be enlightening to compare a method that stacks multiple models versus SPM, which is model-free, and CH-adaptive (CHA), which relies on one model adapting to the intrinsic network structure. Using the 550 networks of Ghasemian et al 1 , we report the results according to mean performance (Figure 1) and win rate (Figure 2) for Precision, AUC-PR and AUC-ROC, following evaluation strategies previously employed 2,7 . Brute-force stacking of algorithms by AI does not perform better than (and is often significantly outperformed by) SPM and one simple brain-bioinspired rule such as CH.…”

mentioning

confidence: 99%

Short Note on Comparing Stacking Modelling Versus Cannistraci-Hebb Adaptive Network Automata for Link Prediction in Complex Networks

Muscoloni¹,

Cannistraci²

2021

Preprint

View full text Add to dashboard Cite

Link prediction is an iconic problem in complex networks because deals with the ability to predict nonobserved existing or future parts of the network structure. The impact of this prediction on real applications can be disruptive: from prediction of covert links between terrorists in their social networks to repositioning of drugs in molecular diseasome networks. Here we compare: (1) an ensemble meta-learning method (Ghasemian et al.), which uses an artificial intelligence (AI) stacking strategy to create a single meta-model from hundreds of other models; (2) a structural predictability method (SPM, Lü et al.), which relies on a theory derived from quantum mechanics and does not assume any model; (3) a modelling rule named Cannistraci-Hebb (CH, Muscoloni et al.), which relies on one brain-bioinspired model adapting to the intrinsic network structure.We conclude that brute-force stacking of algorithms by AI does not perform better than (and is often significantly outperformed by) SPM and one simple brain-bioinspired rule such as CH. This agrees with the Gödel incompleteness: stacking is optimal but incomplete, you cannot squeeze out more than what is already in your features. Hence, we should also pursue AI that resembles human-like physical ‘understanding’ of simple generalized rules associated to complexity. The future might be populated by AI that ‘steals for us the fire from Gods’, towards machine intelligence that creates new rules rather than stacking the ones already known.

show abstract

“…We might expect a network with organized structure to have groups of vertices which are more similar within a group but less similar between groups. Many methods have been explored to address the question of vertex similarity, some directly using methods such as kernels (Jaccard, 1901;Salton & McGill, 1983;Leicht et al, 2006;Kondor & Lafferty, 2002;Smola & Kondor, 2003;Cooper & Barahona, 2010;Fouss et al, 2006), and some indirectly via graph distances or embedding techniques such as graph Laplacian embedding (Lenart, 1998;Fiedler, 1989;Chan et al, 1994;Shi & Malik, 2000;Meila & Shi, 2001;Perrault-joncas & Meila, 2011;Luo et al, 2009;Fouss et al, 2007;Bai et al, 2005;Ghawalby & Hancock, 2015;Huang et al, 2015;Cheng et al, 2019), or inference approaches such as link prediction (Liben-Nowell & Kleinberg, 2007;Zhou et al, 2009;Pech et al, 2019;Zhou et al, 2021). Indeed, the position of vertices in networks can be generalized into approaches that identify deeper mathematical properties related to vertex configurations as in (Brandes, 2016), which provides a general framework for the theoretical idea of vertex similarity.…”

Section: Introductionmentioning

confidence: 99%

Diffusion profile embedding as a basis for graph vertex similarity

Payne¹,

Fuller

Spirou³

et al. 2021

Net Sci

View full text Add to dashboard Cite

We describe here a notion of diffusion similarity, a method for defining similarity between vertices in a given graph using the properties of random walks on the graph to model the relationships between vertices. Using the approach of graph vertex embedding, we characterize a vertex vi by considering two types of diffusion patterns: the ways in which random walks emanate from the vertex vi to the remaining graph and how they converge to the vertex vi from the graph. We define the similarity of two vertices vi and vj as the average of the cosine similarity of the vectors characterizing vi and vj. We obtain these vectors by modifying the solution to a differential equation describing a type of continuous time random walk.This method can be applied to any dataset that can be assigned a graph structure that is weighted or unweighted, directed or undirected. It can be used to represent similarity of vertices within community structures of a network while at the same time representing similarity of vertices within layered substructures (e.g., bipartite subgraphs) of the network. To validate the performance of our method, we apply it to synthetic data as well as the neural connectome of the C. elegans worm and a connectome of neurons in the mouse retina. A tool developed to characterize the accuracy of the similarity values in detecting community structures, the uncertainty index, is introduced in this paper as a measure of the quality of similarity methods.

show abstract

Experimental analyses on 2-hop-based and 3-hop-based link prediction algorithms

Cited by 48 publications

References 55 publications

Progresses and challenges in link prediction

Progresses and challenges in link prediction

Short Note on Comparing Stacking Modelling Versus Cannistraci-Hebb Adaptive Network Automata for Link Prediction in Complex Networks

Diffusion profile embedding as a basis for graph vertex similarity

Contact Info

Product

Resources

About