Causality
# Resources
- To Build Truly Intelligent Machines, Teach Them Cause and Effect
- Representing uncertain knowledge
- Reasoning over time
- Making decisions
- Causal Analysis Introduction - Examples in Python and PyMC
- Granger causality
- The Granger causality test is a statistical hypothesis test for determining whether one time series is useful in forecasting another, first proposed in 1969
- Granger causality is a fundamental technique for causal inference in time series data, commonly used in the social and biological sciences
- PCMCI:
- https://jakobrunge.github.io/tigramite/#tigramite-pcmci-pcmci
- PCMCI causal discovery for time series datasets. It is a 2-step causal discovery method for large-scale time series datasets. The first step is a condition-selection followed by the MCI conditional independence test.
# Talks
- #TALK Interview - Causal Reasoning, Counterfactuals, and the Path to AGI (Judea Pearl)
- #TALK Causality, Bernhard Schölkopf and Stefan Bauer, MLSS 2020:
- #TALK Yoshua Bengio Guest Talk - Towards Causal Representation Learning
# Code
- #CODE Causalai (Salesforce) - A Fast and Scalable framework for Causal Analysis of Time Series and Tabular Data
- #CODE Causalml
- #CODE Causality - Tools for causal analysis
- #CODE CausalImpact (for R)
- #CODE tfcausalimpact - Google’s Causal Impact Algorithm Implemented on Top of TensorFlow Probability
# References
- #PAPER Causal inference with multiple time series: principles and problems (2013)
- #PAPER Towards a Learning Theory of Cause-Effect Inference (Lopez-Paz 2015)
- #PAPER Unsupervised Discovery of El Nino Using Causal Feature Learning on Microlevel Climate Data (Chalupka 2016)
- #PAPER Comparative Benchmarking of Causal Discovery Techniques (Singh 2017)
- #PAPER A Physics-Based Approach to Unsupervised Discovery of Coherent Structures in Spatiotemporal Systems (Rupe 2017)
- #PAPER A Primer on Causal Analysis (2018)
- #PAPER DAGs with NO TEARS: Continuous Optimization for Structure Learning (Zheng 2018)
- #PAPER Local causal states and discrete coherent structures (Rupe 2018)
- #PAPER Learning Functional Causal Models with Generative Neural Networks (Goudet 2018)
- #PAPER Variable-lag Granger Causality for Time Series Analysis (2019)
- #PAPER Learning Sparse Nonparametric DAGs (Zheng 2020)
- #PAPER
When causal inference meets deep learning (Luo 2020)
- Learning causal relations, rather than correlations, is a fundamental problem in both statistical Machine Learning and computer sciences
- Bayesian networks (BNs) can capture causal relations, but learning such a network from data is NP-hard
- Recent work has made it possible to approximate this problem as a continuous optimization task that can be solved efficiently with well-established numerical techniques
- BNs encode the conditional independencies between variables using directed acyclic graphs (DAGs)
- #PAPER
Spacetime Autoencoders Using Local Causal States (Rupe 2020)
- #CODE https://github.com/adamrupe/spacetime_autoencoders
- Local causal states are latent representations that capture organized pattern and structure in complex spatiotemporal systems
- We expand their functionality, framing them as space-time autoencoders
- #PAPER Algorithms for Causal Reasoning in Probability Trees (Genewein 2020)
- #PAPER Off-the-shelf deep learning is not enough, and requires parsimony, Bayesianity, and causality (Vasudevan 2021)
- #PAPER Towards Causal Representation Learning (Schölkopf 2021)
- #PAPER Salesforce CausalAI Library: A Fast and Scalable Framework for Causal Analysis of Time Series and Tabular Data (Arpit 2023)