MNL: A Highly-Efficient Model for Large-scale Dynamic Weighted Directed Network Representation
Minzhi Chen, Chunlin He, Xin Luo
Abstract
A Non-negative Latent-factorization-of-tensors model relying on a <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">N</u> onnegative and <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">M</u> ultiplicative <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">U</u> pdate on <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">I</u> ncomplete <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T</u> ensors (NMU-IT) algorithm facilitates efficient representation learning to a <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">D</u> ynamic <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">W</u> eighted <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">D</u> irected <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">N</u> etwork (DWDN). However, a NMU-IT algorithm leads to slow model convergence and inefficient selection of hyper-parameters. Aiming to address these challenging issues, this work proposes a Momentum-incorporated Biased Non-negative and Adaptive Latent-factorization-of-tensors (MNL) model. It adopts two-fold ideas: 1) incorporating a generalized momentum method into the NMU-IT algorithm to enable fast model convergence; 2) facilitating hyper-parameter slef-adaptation via Particle Swarm Optimization. Empirical studies on four real DWDNs indicate that the proposed MNL is superior to state-of-the-art models in performing efficient representation learning to a DWDN, which is definitely supported by its high computational efficiency and prediction accuracy for missing links of a DWDN. Moreover, its hyper-parameter-free training enables its high practicability in real scenes.