scholarly journals Learning temporal attention in dynamic graphs with bilinear interactions

PLoS ONE ◽  
2021 ◽  
Vol 16 (3) ◽  
pp. e0247936
Author(s):  
Boris Knyazev ◽  
Carolyn Augusta ◽  
Graham W. Taylor
Keyword(s):  
Link Prediction ◽  
Point Processes ◽  
Superior Performance ◽  
Dynamic Graphs ◽  
Temporal Attention ◽  
Short Term ◽  
Baseline Model ◽  
Bilinear Transformation ◽  
Structural Connections

Reasoning about graphs evolving over time is a challenging concept in many domains, such as bioinformatics, physics, and social networks. We consider a common case in which edges can be short term interactions (e.g., messaging) or long term structural connections (e.g., friendship). In practice, long term edges are often specified by humans. Human-specified edges can be both expensive to produce and suboptimal for the downstream task. To alleviate these issues, we propose a model based on temporal point processes and variational autoencoders that learns to infer temporal attention between nodes by observing node communication. As temporal attention drives between-node feature propagation, using the dynamics of node interactions to learn this key component provides more flexibility while simultaneously avoiding issues associated with human-specified edges. We also propose a bilinear transformation layer for pairs of node features instead of concatenation, typically used in prior work, and demonstrate its superior performance in all cases. In experiments on two datasets in the dynamic link prediction task, our model often outperforms the baseline model that requires a human-specified graph. Moreover, our learned attention is semantically interpretable and infers connections similar to actual graphs.

scholarly journals A Survey on Embedding Dynamic Graphs

2023 ◽  
Vol 55 (1) ◽  
pp. 1-37
Author(s):  
Claudio D. T. Barros ◽  
Matheus R. F. Mendonça ◽  
Alex B. Vieira ◽  
Artur Ziviani
Keyword(s):  
Link Prediction ◽  
Point Processes ◽  
Graph Embedding ◽  
Graph Visualization ◽  
Dynamic Graphs ◽  
Dynamic Graph ◽  
Topological Evolution ◽  
Evolution Feature ◽  
Feature Evolution ◽  
And Diffusion

Embedding static graphs in low-dimensional vector spaces plays a key role in network analytics and inference, supporting applications like node classification, link prediction, and graph visualization. However, many real-world networks present dynamic behavior, including topological evolution, feature evolution, and diffusion. Therefore, several methods for embedding dynamic graphs have been proposed to learn network representations over time, facing novel challenges, such as time-domain modeling, temporal features to be captured, and the temporal granularity to be embedded. In this survey, we overview dynamic graph embedding, discussing its fundamentals and the recent advances developed so far. We introduce the formal definition of dynamic graph embedding, focusing on the problem setting and introducing a novel taxonomy for dynamic graph embedding input and output. We further explore different dynamic behaviors that may be encompassed by embeddings, classifying by topological evolution, feature evolution, and processes on networks. Afterward, we describe existing techniques and propose a taxonomy for dynamic graph embedding techniques based on algorithmic approaches, from matrix and tensor factorization to deep learning, random walks, and temporal point processes. We also elucidate main applications, including dynamic link prediction, anomaly detection, and diffusion prediction, and we further state some promising research directions in the area.

ACHIEVING SUPERIOR INTERNATIONALIZATION THROUGH STRATEGIC AMBIDEXTERITY

2007 ◽  
Vol 15 (01) ◽  
pp. 43-77 ◽  
Author(s):  
MARY HAN
Keyword(s):  
Multinational Enterprise ◽  
Field Research ◽  
New Venture ◽  
Superior Performance ◽  
International New Venture ◽  
Short Term ◽  
Firm Level ◽  
Optimal Approach ◽  
Level Performance

In this paper, I examine the optimal approach to internationalization. Drawing from the ambidexterity literature, I build on the concept of structural ambidexterity and suggest that the optimal approach occurs when firms pursue 'strategic ambidexterity,' which is characterized by simultaneously executing paradoxical strategies of pro-profit and pro-growth. I examine this approach through empirical study based on in-depth field research of two cases: Merrill Lynch, a multinational enterprise and Comdirect Bank, an international new venture. I find that a) firms that pursue strategic ambidexterity in their internationalization effort achieve above-average internationalization performance in the short term; and b) firms that pursue strategic ambidexterity in their internationalization effort also achieve above-average firm-level performance in the long term. I conclude that strategic ambidexterity is the optimal strategy by which to achieve superior performance.

Entry Timing as a Mixed Gamble in Cross-border Acquisition Waves: A study of Family Firms

2021 ◽  
pp. 089448652110261
Author(s):  
Mohammad Fuad ◽  
Vinod Thakur ◽  
Ashutosh Kumar Sinha
Keyword(s):  
Family Firms ◽  
Superior Performance ◽  
Short Term ◽  
Family Control ◽  
Cross Border ◽  
Trade Offs ◽  
Entry Timing ◽  
The Relationship ◽  
Financial Losses

We draw upon the mixed gamble perspective to investigate the entry timing decisions made by family firms in the context of cross-border acquisition (CBA) waves. We argue that family-controlled firms trade-off short-term SEW and financial losses in favor of long-term SEW and financial gains, while moving early in CBA waves. Findings suggest that family-controlled firms have a higher preference for early movement compared with nonfamily-controlled firms. Further, we show that founder’s presence on the board and acquirer’s superior performance amplifies the mixed gamble trade-offs, thereby strengthening the relationship between family control and early movement within CBA waves.

scholarly journals AddGraph: Anomaly Detection in Dynamic Graph Using Attention-based Temporal GCN

2019 ◽  
Author(s):  
Li Zheng ◽  
Zhenpeng Li ◽  
Jian Li ◽  
Zhao Li ◽  
Jun Gao
Keyword(s):  
Anomaly Detection ◽  
Dynamic Graphs ◽  
Dynamic Graph ◽  
Short Term ◽  
Convolutional Network ◽  
Attention Model ◽  
Temporal Features ◽  
Structural Content ◽  
Real World Datasets

Anomaly detection in dynamic graphs becomes very critical in many different application scenarios, e.g., recommender systems, while it also raises huge challenges due to the high flexible nature of anomaly and lack of sufficient labelled data. It is better to learn the anomaly patterns by considering all possible features including the structural, content and temporal features, rather than utilizing heuristic rules over the partial features. In this paper, we propose AddGraph, a general end-to-end anomalous edge detection framework using an extended temporal GCN (Graph Convolutional Network) with an attention model, which can capture both long-term patterns and the short-term patterns in dynamic graphs. In order to cope with insufficient explicit labelled data, we employ the negative sampling and margin loss in training of AddGraph in a semi-supervised fashion. We conduct extensive experiments on real-world datasets, and illustrate that AddGraph can outperform the state-of-the-art competitors in anomaly detection significantly.

scholarly journals On Simulating the Propagation and Countermeasures of Hate Speech in Social Networks

2021 ◽  
Vol 11 (24) ◽  
pp. 12003
Author(s):  
Maite Lopez-Sanchez ◽  
Arthur Müller
Keyword(s):  
Social Networks ◽  
Hate Speech ◽  
Research Literature ◽  
Short Term ◽  
Baseline Model ◽  
Agent Based ◽  
Online Social Media ◽  
Prejudice And Discrimination ◽  
Positive Effect

Hate speech expresses prejudice and discrimination based on actual or perceived innate characteristics such as gender, race, religion, ethnicity, colour, national origin, disability or sexual orientation. Research has proven that the amount of hateful messages increases inevitably on online social media. Although hate propagators constitute a tiny minority—with less than 1% participants—they create an unproportionally high amount of hate motivated content. Thus, if not countered properly, hate speech can propagate through the whole society. In this paper we apply agent-based modelling to reproduce how the hate speech phenomenon spreads within social networks. We reuse insights from the research literature to construct and validate a baseline model for the propagation of hate speech. From this, three countermeasures are modelled and simulated to investigate their effectiveness in containing the spread of hatred: Education, deferring hateful content, and cyber activism. Our simulations suggest that: (1) Education consititutes a very successful countermeasure, but it is long term and still cannot eliminate hatred completely; (2) Deferring hateful content has a similar—although lower—positive effect than education, and it has the advantage of being a short-term countermeasure; (3) In our simulations, extreme cyber activism against hatred shows the poorest performance as a countermeasure, since it seems to increase the likelihood of resulting in highly polarised societies.

An Evaluation of Experimental How-Malfunctioned Codes

1970 ◽  
Vol 12 (3) ◽  
pp. 261-269
Author(s):  
Nilss M. Aume ◽  
Donald A. Topmiller
Keyword(s):  
Error Rate ◽  
Superior Performance ◽  
Average Error ◽  
Short Term ◽  
Experimental Conditions ◽  
Associate Learning ◽  
High Association ◽  
Maintenance Technicians ◽  
Force Maintenance

In three learning-recall studies, the existing coding procedure by which Air Force maintenance technicians record and describe equipment malfunctions was compared to four experimental codes. The four three-symbol experimental codes included all numeric symbols, all alphabetical symbols, alphanumeric symbols, and a mnemonic code that had a high association value with the corresponding descriptor statements. Performance using the five experimental conditions was evaluated in three separate studies: first, under short-term recall, second, under paired-associate learning, and third, under long-term recall. All experiments demonstrated superior performance with the mnemonic code condition. The mnemonic codes produced approximately one-half the average error rate for short-term recall, were twice as easily learned, and achieved nearly half the error rate for long-term recall when compared to the existing coding technique.

CEO Turnover, Network Effects, and Firm Performance

2020 ◽  
Vol 10 (2) ◽  
pp. 54-72
Author(s):  
Yanan Sun ◽  
Peiqin Zhang ◽  
David Wierschem ◽  
Francis A Mendez Mediavilla
Keyword(s):  
Firm Performance ◽  
Network Effects ◽  
Generalized Method Of Moments ◽  
Market Performance ◽  
Ceo Turnover ◽  
Superior Performance ◽  
Short Term ◽  
Moderating Role

This article applies network and organizational theory to examine the effect of CEO turnover on firm accounting and market performance in both short-term and long-term. In addition, this research investigates the moderating role of network effects using cluster analysis. Using a system generalized method of moments (GMM) estimation of panel data obtained from Compustat and S&P's Execucomp database, this study finds that it is less likely to have superior performance in the long-term for firms with frequent CEO turnover. While it is more likely to have better accounting performance over the short-term, but less likely to have superior market performance. This study further validates the moderating role of network effects. This article contributes to the research by providing new insights of CEO turnover effects on firm performance and investigating the moderation effect of network structure. The findings also provide practical suggestions for firms that experience frequent changes of their CEOs.

Conceptual short-term memory (CSTM) supports core claims of Christiansen and Chater

2016 ◽  
Vol 39 ◽  
Author(s):  
Mary C. Potter
Keyword(s):  
Working Memory ◽  
Rapid Serial Visual Presentation ◽  
Language Comprehension ◽  
Short Term Memory ◽  
Visual Presentation ◽  
Long Term Memory ◽  
Short Term ◽  
Term Memory ◽  
Use Of Knowledge

AbstractRapid serial visual presentation (RSVP) of words or pictured scenes provides evidence for a large-capacity conceptual short-term memory (CSTM) that momentarily provides rich associated material from long-term memory, permitting rapid chunking (Potter 1993; 2009; 2012). In perception of scenes as well as language comprehension, we make use of knowledge that briefly exceeds the supposed limits of working memory.

Ultrastructural investigations of MDL 19,660-induced effects on the canine platelet and megakaryocyte

1990 ◽  
Vol 48 (3) ◽  
pp. 374-375
Author(s):  
D.E. Loudy ◽  
J. Sprinkle-Cavallo ◽  
J.T. Yarrington ◽  
F.Y. Thompson ◽  
J.P. Gibson
Keyword(s):  
Antidepressant Drug ◽  
Beagle Dogs ◽  
Long Term Effects ◽  
Short Term ◽  
Platelet Counts ◽  
Toxicological Studies ◽  
Induced Aggregation ◽  
Internal Architecture

Previous short term toxicological studies of one to two weeks duration have demonstrated that MDL 19,660 (5-(4-chlorophenyl)-2,4-dihydro-2,4-dimethyl-3Hl, 2,4-triazole-3-thione), an antidepressant drug, causes a dose-related thrombocytopenia in dogs. Platelet counts started to decline after two days of dosing with 30 mg/kg/day and continued to decrease to their lowest levels by 5-7 days. The loss in platelets was primarily of the small discoid subpopulation. In vitro studies have also indicated that MDL 19,660: does not spontaneously aggregate canine platelets and has moderate antiaggregating properties by inhibiting ADP-induced aggregation. The objectives of the present investigation of MDL 19,660 were to evaluate ultrastructurally long term effects on platelet internal architecture and changes in subpopulations of platelets and megakaryocytes.Nine male and nine female beagle dogs were divided equally into three groups and were administered orally 0, 15, or 30 mg/kg/day of MDL 19,660 for three months. Compared to a control platelet range of 353,000- 452,000/μl, a doserelated thrombocytopenia reached a maximum severity of an average of 135,000/μl for the 15 mg/kg/day dogs after two weeks and 81,000/μl for the 30 mg/kg/day dogs after one week.

Comparison of Auditory, Language, Memory, and Attention Abilities in Children With and Without Listening Difficulties

2020 ◽  
Vol 29 (4) ◽  
pp. 710-727
Author(s):  
Beula M. Magimairaj ◽  
Naveen K. Nagaraj ◽  
Alexander V. Sergeev ◽  
Natalie J. Benafield
Keyword(s):  
Auditory Processing ◽  
Short Term Memory ◽  
Group Differences ◽  
Long Term Memory ◽  
Short Term ◽  
Significant Group ◽  
Term Memory ◽  
Memory And Attention ◽  
Speed And Accuracy

Objectives School-age children with and without parent-reported listening difficulties (LiD) were compared on auditory processing, language, memory, and attention abilities. The objective was to extend what is known so far in the literature about children with LiD by using multiple measures and selective novel measures across the above areas. Design Twenty-six children who were reported by their parents as having LiD and 26 age-matched typically developing children completed clinical tests of auditory processing and multiple measures of language, attention, and memory. All children had normal-range pure-tone hearing thresholds bilaterally. Group differences were examined. Results In addition to significantly poorer speech-perception-in-noise scores, children with LiD had reduced speed and accuracy of word retrieval from long-term memory, poorer short-term memory, sentence recall, and inferencing ability. Statistically significant group differences were of moderate effect size; however, standard test scores of children with LiD were not clinically poor. No statistically significant group differences were observed in attention, working memory capacity, vocabulary, and nonverbal IQ. Conclusions Mild signal-to-noise ratio loss, as reflected by the group mean of children with LiD, supported the children's functional listening problems. In addition, children's relative weakness in select areas of language performance, short-term memory, and long-term memory lexical retrieval speed and accuracy added to previous research on evidence-based areas that need to be evaluated in children with LiD who almost always have heterogenous profiles. Importantly, the functional difficulties faced by children with LiD in relation to their test results indicated, to some extent, that commonly used assessments may not be adequately capturing the children's listening challenges. Supplemental Material https://doi.org/10.23641/asha.12808607

Sign in / Sign up

Export Citation Format

Share Document