R2.1 Machine Learning Approaches

25 Sep 2018
10:30 - 12:30
lecture Hall 4

R2.1 Machine Learning Approaches

R2.1 _ Understanding species distribution, population dynamics and phenology by machine learning. (Session Chairs: Antonino Staiano, Alberto Lanzoni and Friedrich Recknagel)

Inferential modelling by machine learning techniques allows efficient and user-friendly analysis and synthesis of highly complex ecological data. Methods like random forest, quantile regression forest, Maxent, GARP have been successfully applied for species distribution modelling of extensive spatial data resulting in species response curves that describe a species’ response to a given habitat condition, and illustrate specific habitat requirements for the species. Non-supervised artificial neural networks, regression trees and canonical correspondence analyses allow to ordinate and classify complex ecological data. Applications of support vector machines, supervised artificial neural networks and evolutionary algorithms to large spatial and temporal data allow to predict population dynamics and reveal phenology.

This session welcomes papers on all aspects of inferential modelling of ecological data by means of novel machine learning techniques.

Talk 01-07

A machine learning approach to the assessment of the vulnerability of Posidonia oceanica meadows
Elena Catucci and Michele Scardi

Causal relationships of Cylindrospermopsis dynamics with water temperature and N/P-ratios: a meta-analysis across lakes with different climate based on inferential modelling by HEA
Friedrich Recknagel, Tamar Zohary, Ilia Ostrovsky, Jacqueline Rücker, Philip Orr, Christina Castello Branco, Brigitte Nixdorf and Ricardo Tezini

Dynamics of four cyanobacteria in the Nakdong River, South Korea over 24 years (1993-2016) patternized by an artificial neural network
Hyo Gyeom Kim, Sungwon Hong, Dong-Kyun Kim and Gea-Jae Joo

Modeling Green Peach Aphid populations exposed to elicitors inducing plant resistance on peach
Alberto Lanzoni, Francesco Camastra, Angelo Ciaramella, Antonino Staiano and Giovanni Burgio

Modelling urban bird breeding sites with a random forest classifier using indicators of spatial heterogeneity in plant communities derived from earth observation data
Thilo Wellmann, Angela Lausch, Sebastian Scheuer and Dagmar Haase

Supervised learning methods to predict species interactions based on traits and phylogeny
Michiel Stock and Bernard De Baets

Integrating context-based recommendation with deep CNN image classification for on-site plant species identification
Hans Christian Wittich, David Boho, and Patrick Mäder