Code for NeurIPS 2021 paper 'Spatio-Temporal Variational Gaussian Processes'

Last update: Sep 16, 2022

Overview

Spatio-Temporal Variational GPs

This repository is the official implementation of the methods in the publication:

O. Hamelijnck, W.J. Wilkinson, N.A. Loppi, A. Solin, and T. Damoulas (2021). Spatio-temporal variational Gaussian processes. In Neural Information Processing Systems (NeurIPS). [arXiv]

Citing this work:

@inproceedings{hamelijnck2021spatio,
	title={Spatio-Temporal Variational {G}aussian Processes},
	author={Hamelijnck, Oliver and Wilkinson, William and Loppi, Niki and Solin, Arno and Damoulas, Theodoros},
	booktitle={Advances in Neural Information Processing Systems (NeurIPS)},
	year={2021},
}

Experiment Setup

This has been tested on a Macbook Pro. All spatio-temporal VGP models have been implemented within the Bayes-Newton package.

Environment Setup

We recommend using conda:

conda create -n spatio_gp python=3.7
conda activate spatio_gp

Then install the required python packages:

pip install -r requirements.txt

Data Download

Pre-processed Data

All data, preprocessed and split into train-test splits used in the paper is provided at https://doi.org/10.5281/zenodo.4531304. Download the folder and place the corresponding datasets into experiments/*/data folders.

Manual Data Setup

We also provide scripts to generate the data manually:

make data

which will download the relevant London air quality and NYC data, clean them, and split into train-test splits.

Running Experiments

To run all experiments across all training folds run:

make experiments

To run an individual experiment refer to the Makefile.

Baselines used

GPFlow2 : https://github.com/GPflow/GPflow
GPYTorch: https://github.com/cornellius-gp/gpytorch

License

This software is provided under the MIT license.

A Tensorflow based library for Time Series Modelling with Gaussian Processes

Markovflow Documentation | Tutorials | API reference | Slack What does Markovflow do? Markovflow is a Python library for time-series analysis via prob

24 Dec 12, 2022

Deep generative modeling for time-stamped heterogeneous data, enabling high-fidelity models for a large variety of spatio-temporal domains.

Neural Spatio-Temporal Point Processes [arxiv] Ricky T. Q. Chen, Brandon Amos, Maximilian Nickel Abstract. We propose a new class of parameterizations

75 Dec 19, 2022

Learning Spatio-Temporal Transformer for Visual Tracking

Comments

Typo: `discretegrid` function expects `numpy` being imported as `nnp` instead of `np`

Probably not being used, but discretegrid function expects numpy being imported as nnp instead of np

https://github.com/AaltoML/spatio-temporal-GPs/blob/800428a239e9fb91b3472f806241e5396273789b/experiments/utils.py#L217..L236

opened by diegovalenzuelaiturra 0

Code for NeurIPS 2021 paper 'Spatio-Temporal Variational Gaussian Processes'

Related tags

Overview

Spatio-Temporal Variational GPs

Citing this work:

Experiment Setup

Environment Setup

Data Download

Pre-processed Data

Manual Data Setup

Running Experiments

Baselines used

License

You might also like...

A Tensorflow based library for Time Series Modelling with Gaussian Processes

Deep generative modeling for time-stamped heterogeneous data, enabling high-fidelity models for a large variety of spatio-temporal domains.

Learning Spatio-Temporal Transformer for Visual Tracking

Spontaneous Facial Micro Expression Recognition using 3D Spatio-Temporal Convolutional Neural Networks

Digital Twin Mobility Profiling: A Spatio-Temporal Graph Learning Approach

DeepSTD: Mining Spatio-temporal Disturbances of Multiple Context Factors for Citywide Traffic Flow Prediction

Spatio-Temporal Entropy Model (STEM) for end-to-end leaned video compression.

Self-supervised spatio-spectro-temporal represenation learning for EEG analysis

[CVPR 2022 Oral] TubeDETR: Spatio-Temporal Video Grounding with Transformers

Comments

Typo: `discretegrid` function expects `numpy` being imported as `nnp` instead of `np`

Owner

AaltoML

Code for the paper "Spatio-temporal Self-Supervised Representation Learning for 3D Point Clouds" (ICCV 2021)

Implementation of "Fast and Flexible Temporal Point Processes with Triangular Maps" (Oral @ NeurIPS 2020)

This repository contains the data and code for the paper "Diverse Text Generation via Variational Encoder-Decoder Models with Gaussian Process Priors" (SPNLP@ACL2022)

ESTDepth: Multi-view Depth Estimation using Epipolar Spatio-Temporal Networks (CVPR 2021)

Implementation of the "PSTNet: Point Spatio-Temporal Convolution on Point Cloud Sequences" paper.

Implementation of the "Point 4D Transformer Networks for Spatio-Temporal Modeling in Point Cloud Videos" paper.

Codes for TIM2021 paper "Anchor-Based Spatio-Temporal Attention 3-D Convolutional Networks for Dynamic 3-D Point Cloud Sequences"

Code to reproduce the experiments from our NeurIPS 2021 paper " The Limitations of Large Width in Neural Networks: A Deep Gaussian Process Perspective"

A highly efficient and modular implementation of Gaussian Processes in PyTorch

A Python implementation of global optimization with gaussian processes.