- Tweets 58
- Following 410
- Followers 209
- Likes 783
Maximilian Pichler retweeted
We are hiring: 4 PhD Positions in Ecology / Data Science and one Ecological Data Scientist. For more information, see uni-regensburg.de/universita…
1
5
8
843
#PhD position!
I’m looking for a PhD Student (1+3 years) to model insect populations using statistical models + deep learning at #TheoreticalEcology, Regensburg (Germany).
Join our team - please RT!
Details: karriere.uni-regensburg.de/t…
#Insects #DeepLearning #Ecology #Monitoring
26
93
5,415
In our new preprint, “Inferring processes within dynamic forest models using hybrid modeling” @KaberYannek I present a new hybrid modeling approach for jointly calibrating a DVM with embedded #deepneuralnetwork arxiv.org/abs/2508.01228 1/4
#deeplearning #forestdynamics
1
144
We introduce forest-informed neural networks (FINNs), a new DVM in which processes can be replaced by #DNN and the entire model is calibrated jointly. FINN can approximate the functional shapes of otherwise misspecified processes and achieve better predictive accuracy 3/4
1
137
Moreover, we can use explainable AI to understand the learned functional form of the replaced process. We demonstrated this using the BCI plot by replacing the growth process with a #DNN. We found plausible dbh-growth and light-growth functions learned by FINN 4/4
105
Maximilian Pichler retweeted
In @nature, Delavaux et al., 2024 suggest that species richness on oceanic islands is reduced by mutualism rates on the mainland. In arxiv.org/abs/2411.15105, @_Max_Pichler and I had another look at this question and conclude that this effect is likely a statistical artefact.
2
10
33
3,273
In our @EcographyJourna software note ‘cito': an R package for training neural networks using ‘torch' (#CRAN #rstats) led by @amesoeder and together with @florianhartig we present a new package for fitting and interpreting #DNN (#DL) doi.org/10.1111/ecog.07143 @TheoEcolUR 1/5
2
21
71
6,605
Moreover, cito is tailored for ecological data, beside typical loss functions we support likelihoods such as Poisson, negative binomial, or multivariate probit link for joint species distribution models #JSDM. 4/5
1
1
256
We have prepared an extensive overview and documentation (especially for beginners) with examples in our vignettes (citoverse.github.io/cito/und… Articles) 5/5:
5
239
Maximilian Pichler retweeted
📊 Exciting news! Prepare to boost your skills and expertise in statistics with our stellar lineup of instructors:
@vdVeenB @MatthewBJane @nj_clark @ucfagls @PhilipLeftwich @florianhartig @_Max_Pichler
Don't miss out on this opportunity 🚀physalia-courses.org/courses…
5
9
2,600
cito v1.1 #rstats package for deep neural networks (#DL #DNN) (with formula syntax) is now available on #CRAN. New features include likelihoods such as the negative binomial distribution and easy hyperparameter tuning: cran.r-project.org/web/packa… 1/4
1
13
40
5,090
For more details and explanations on how to train and interpret DNNs, see our extensive documentation (including #SDM and #MSDM examples) that also covers advanced topics such as custom loss functions and residual checks (under articles on citoverse.github.io/cito/)!
196
Maximilian Pichler retweeted
Come work with us - I am looking to fill a 3-yr position for a statistical postdoc / scientific programmer to continue the development of the DHARMa #Rstats #CRAN package for #glmm residual diagnostics. Full job advertisement is here uni-regensburg.de/assets/bio…
2
60
62
16,306
cito v1.0.2 #rstats package for deep neural networks(#DL #DNN) is now available on #CRAN. New features include confidence intervals and p-values for #xAI metrics: cran.r-project.org/web/packa… @florianhartig @amesoeder
1
9
27
2,709
An important new feature is the baseline loss, which helps in training the #DNN. It allows, for example, to identify inappropriate learning rates:
1
193
To learn more about example workflows such as #SDM, multi-species distribution models, or advanced stuff like quantile regression (e.g. for #conformalPrediction), check out the vignettes: cran.r-project.org/web/packa…
1
1
202