StatEcoNet: Statistical Ecology Neural Networks for Species Distribution Modeling
Authors: Eugene Seo, Rebecca A. Hutchinson, Xiao Fu, Chelsea Li, Tyler A. Hallman, John Kilbride, W. Douglas Robinson513-521
AAAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | The advantages of Stat Eco Net over related approaches are demonstrated on simulated datasets as well as bird species data. We evaluated our model with both simulated and avian point count data. |
| Researcher Affiliation | Academia | 1School of Electrical Engineering and Computer Science, Oregon State University, Corvallis, OR 97331 2Department of Fisheries and Wildlife, Oregon State University, Corvallis, OR 97331 3College of Earth, Ocean, and Atmospheric Sciences, Oregon State University, Corvallis, OR 97331 4Monitoring Department, Swiss Ornithological Institute, Sempach, Switzerland |
| Pseudocode | No | The paper mentions "algorithmic details are in the supplement" and describes the training process, but it does not include a structured pseudocode block or algorithm labeled in the main text. |
| Open Source Code | Yes | The code and supplementary material are available at https://github.com/Hutchinson-Lab/Stat Eco Net-AAAI21. |
| Open Datasets | Yes | We simulated data to evaluate the models ability to predict probabilities and observations as well as discover important features under the assumed model. We analyzed 10,845 5-minute point count bird surveys extracted from the Oregon 2020 dataset collected in Oregon, United States (Robinson et al. 2020). |
| Dataset Splits | Yes | In total, we simulated training and validation sets from the eight combinations of M {100, 1000}, T {3, 10}, and feature-occupancy/detection model {linear, nonlinear}. Testing sets always had M = 1000 for more robust performance estimates. We divided these data into three spatially distinct cross-validation folds (Valavi et al. 2018). |
| Hardware Specification | No | The paper does not specify any hardware details such as GPU models, CPU types, or memory specifications used for running the experiments. |
| Software Dependencies | No | The paper does not provide specific version numbers for any software dependencies or libraries used in the implementation (e.g., Python, PyTorch, TensorFlow versions). |
| Experiment Setup | Yes | For Stat Eco Net, we selected the key parameters, i.e., learning rate, batch size, number of neurons per layer, and number of layers, from {0.01, 0.001, 0.0001}, {32, all}, {8, 16, 32, 64}, and {1, 3}, respectively, to maximize the AUPRC performance on the validation set. For OD-BRT, we used Bayesian optimization (Snoek, Larochelle, and Adams 2012; Yan 2016) to tune the shrinkage, bag fraction, tree depth, and number of trees... |