density_regression

Quick Demo

Run this Google Colab.

or

notebook in density_regression.ipynyb

or

python file (full comparision, install prerequisite packages first to import library):

python demo/run_cubic.py
python demo/density_regression.py

To prepare:

Library

Install prerequisite packages:

pip install -r requirements.txt

Dataset

Download dataset:

bash depth_estimation/download_data.sh

To run experiments:

Time series weather forecasting

python <method_file> --exp_idx=<idx>

where the parameters are the following:

• <method_file>: file stored the code of method. E.g., <method_file> = time_series/density_regression.py
• <idx>: index of experiment. E.g., <idx> = 1

Benchmark UCI

python uci/main.py --datasets=<dataset_name> 

where the parameters are the following:

• <dataset_name>: name of the sub-dataset in UCI. E.g., <dataset_name> = "wine"

Monocular depth estimation

python depth_estimation/main.py --model=<method_name> 

where the parameters are the following:

• <method_name>: name of method. E.g., <method_name> = "densityregressor"

References

Based on code of:

Time series forecasting
TensorFlow.

Evidential Deep Learning
Amini, Alexander and Schwarting, Wilko and Soleimany, Ava and Rus, Daniela.
arXiv:1910.02600.

Methods for comparing uncertainty quantifications for material property predictions
Kevin Tran, Willie Neiswanger, Junwoong Yoon, Eric Xing, Zachary W. Ulissi.
arXiv:1912.10066.

License

This source code is released under the Apache-2.0 license, included here.