A deep stochastical and predictive analysis of users mobility based on Auto-Regressive processes and pairing functions

Fazio, Peppino; Mehić, Miralem; Vozňák, Miroslav

doi:10.1016/j.jnca.2020.102778

Cited by 10 publications

(1 citation statement)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We used the deep multiple regression model to solve the problem of NPP fitting by multiple drivers (Sun et al, 2016;Zhang et al, 2019). The algorithm used in this study was the stacked denoising auto-encoder (SDAE), which was developed based on a denoising auto-encoder (DAE) that could learn the original data's characteristics through encoding and decoding and identify the more complex variation characteristics of the driver factors (Fazio et al, 2020). The SDAE is a typical deep learning model (Figure 4) with a learning process that is divided into two parts: an encoding process and a decoding process.…”

Section: Response Relation Modelmentioning

confidence: 99%

Contrasting response mechanisms and ecological stress of net primary productivity in sub-humid to arid transition regions: a case study from the Loess Plateau, China

Wang,

Zhang,

Chang

et al. 2023

Front. Ecol. Evol.

View full text Add to dashboard Cite

With the intensification of global change, the overall functions and structures of ecosystems in sub-humid to arid transition regions have changed to varying degrees. The Loess Plateau, as a typical case of such regions, plays a great role in the study of net primary productivity (NPP) for estimating the sustainability of the Earth’s carrying capacity in terrestrial ecosystem process monitoring. In the research on contrasting response mechanisms and ecological stress of NPP, the main innovations of this study are as follows. On the basis of the indicator system constructed from satellite imagery and meteorological data, we introduced deep multiple regressive models to reveal the relationship between NPP and the identified driving factors and then creatively proposed ecological stress (ES) evaluation models from the perspective of vegetation productivity. The findings are as follows: 1) From 2000 to 2019, the changes in driving factors presented a clear regional character, and the annual NPP maintained a fluctuating increasing trend (with a value of 4.57 g·m2·a−1). From the perspective of spatial distribution, the growth rate of NPP gradually increased from arid to sub-humid regions. 2) The effects of different driving factors on NPP changes and specific NPPs varied greatly across different regions. Arid and semi-arid regions were mainly controlled by precipitation (20.49%), temperature (15.21%), and other related factors, whereas sub-humid regions were mainly controlled by solar radiation, such as net surface solar radiation (NSSR) (8.71%) and surface effective radiation (SER) (7.93%). The main driving factors of NPP change varied under different soil conditions. 3) The spatio-temporal patterns of NPP approximated those of ES, but the effects of the latter significantly differed across ecological functional regions and land uses. This research on the Loess Plateau can serve as a valuable reference for future research on realizing ecosystem restoration and protection in sub-humid to arid transition regions.

show abstract

Section: Response Relation Modelmentioning

confidence: 99%