Multi-level Multi-task Learning for Modeling Cross-Scale Interactions in Nested Geospatial Data

Yuan, Shuai; Zhou, Jiayu; Tan, Pang‐Ning; Fergus, Emi; Wagner, Tyler; Soranno, Patricia A.

doi:10.1109/icdm.2017.154

Cited by 4 publications

(7 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Spatial attributes often exist at different spatial scales [253]. Even symbolic localization results have this issue.…”

Section: Hierarchical and Multi-scaledmentioning

confidence: 99%

“…• Reinforcement Learning, widely used in sequential decision-making, can deal with the incompleteness [199] and dynamics [104,192,218] of trajectories or spatiotemporal sequences. • Multi-task Learning [67,164,253,273] and Multi-view Learning [260,262,272], which make full use of data for improved overall performance, can contend with scarcity of labels, as well as bias and heterogeneity of data during training. • Transfer Learning [72,245], borrowing labeled data or knowledge from related domains, can deal with limited data availability and bias of data in a certain domain.…”

Section: Technique Facetmentioning

confidence: 99%

“…A variety of decision-making tasks based on SID are relevant, such as the prediction of next location(s) [48,102,104,105,181,262], traffic volume [141,199], and spatiotemporal variables [46,67,147,164,245,253]; the recommendation of POIs [82,155,264,273] or routes [72]; and the 7 Lloyd's algorithm [143] is originally for Voronoi-based iterative centroid estimation. planning of task assignments [192] or site selection [40,272].…”

Section: Decision-making Using Low-quality Sidmentioning

confidence: 99%

“…Scarcity of Labels. This issue has been addressed in unsupervised learning (such as EM [82], AutoEncoder [48,236], and GAN (Generative Adversarial Network) [48]), semi-supervised learning (such as co-training [46] and PU learning [40]), and multi-task learning [67,253,273]. Considering multiple latent temporal parameters in POI recommendation, Hosseini et al [82] retrieve multiaspect temporal similarity maps to reduce user-location matrix sparseness and use the EM algorithm to compensate for incomplete data at each temporal scale.…”

Section: Decision-making Using Low-quality Sidmentioning

confidence: 99%

See 3 more Smart Citations

Spatial Data Quality in the Internet of Things: Management, Exploitation, and Prospects

Jensen

et al. 2022

ACM Comput. Surv.

View full text Add to dashboard Cite

With the continued deployment of the Internet of Things (IoT) , increasing volumes of devices are being deployed that emit massive spatially referenced data. Due in part to the dynamic, decentralized, and heterogeneous architecture of the IoT, the varying and often low quality of spatial IoT data (SID) presents challenges to applications built on top of this data. This survey aims to provide unique insight to practitioners who intend to develop IoT-enabled applications and to researchers who wish to conduct research that relates to data quality in the IoT setting. The survey offers an inventory analysis of major data quality dimensions in SID and covers significant data characteristics and associated quality considerations. The survey summarizes data quality related technologies from both task and technique perspectives. Organizing the technologies from the task perspective, it covers recent progress in SID quality management, encompassing location refinement, uncertainty elimination, outlier removal, fault correction, data integration, and data reduction; and it covers low-quality SID exploitation, encompassing querying, analysis, and decision-making techniques. Finally, the survey covers emerging trends and open issues concerning the quality of SID.

show abstract

“…Spatial attributes often exist at different spatial scales [253]. Even symbolic localization results have this issue.…”

Section: Hierarchical and Multi-scaledmentioning

confidence: 99%

Section: Technique Facetmentioning

confidence: 99%

Section: Decision-making Using Low-quality Sidmentioning

confidence: 99%

Section: Decision-making Using Low-quality Sidmentioning

confidence: 99%

See 2 more Smart Citations

Spatial Data Quality in the Internet of Things: Management, Exploitation, and Prospects

Jensen

et al. 2022

ACM Comput. Surv.

View full text Add to dashboard Cite

show abstract

“…Other aspects of crypto network security could also be studied. Machine learning techniques can be applied to predict the network attack probabilities in the future [20][21][22][23][24].…”

Section: Discussionmentioning

confidence: 99%

A review of crypto networks

Zhang

2018

Preprint

View full text Add to dashboard Cite

Bitcoin is a crypto currency system that has been rapidly adopted due to its anonymity and decentralized properties. Blockchain is the underpinning technology that maintains the Bitcoin transaction ledger. The blockchain network operates in a state of consensus, which automatically checks in with itself periodically. One of biggest innovation by Bitcoin system is that it is a new way to develop open networks. Anything that happens on the blockchain is a function of the network as a whole. Crypto networks represent a fundamental shift in the way our society transacts, organizes, and works with each other, which could be explained and deciphered by econophysics of the network itself. From a networking perspective, we reviewed a list of current literatures that studied the crypto networks, mostly Bitcoin transaction networks. We identified the potential research areas that would further provide insights into the design of a more resilient and secure crypto networks.

show abstract

Spatio-Temporal Multi-Task Learning via Tensor Decomposition

Zhou

Tan

et al. 2021

IEEE Trans. Knowl. Data Eng.

View full text Add to dashboard Cite

Multi-level Multi-task Learning for Modeling Cross-Scale Interactions in Nested Geospatial Data

Cited by 4 publications

References 9 publications

Spatial Data Quality in the Internet of Things: Management, Exploitation, and Prospects

Spatial Data Quality in the Internet of Things: Management, Exploitation, and Prospects

A review of crypto networks

Spatio-Temporal Multi-Task Learning via Tensor Decomposition

Contact Info

Product

Resources

About