Kanvaly Fadiga
Abstract
Smart homes are Cyber-Physical Systems (CPS) where multiple devices and controllers cooperate to achieve high-level goals. Causal knowledge on relations between system entities is essential for enabling system self-adaption to dynamic changes. As house configurations are diverse, this knowledge is difficult to obtain. In previous work, we proposed to generate Causal Bayesian Networks (CBN) as follows. Starting with considering all possible relations, we progressively discarded non-correlated variables. Next, we identified causal relations from the remaining correlations by employing “do-operations.” The obtained CBN could then be employed for causal inference. The main challenges of this approach included “non-doable variables” and limited scalability. To address these issues, we propose three extensions: (i) early pruning weakly correlated relations to reduce the number of required do-operations, (ii) introducing aggregate variables that summarize relations between weakly coupled sub-systems, and (iii) applying the method a second time to perform indirect do interventions and handle non-doable relations. We illustrate and evaluate the efficiency of these contributions via examples from the smart home and power grid domain. Our proposal leads to a decrease in the number of operations required to learn the CBN and in an increased accuracy of the learned CBN, paving the way toward applications in large CPS.
- [1] [n. d.]. iCasa simulator Web page.Google Scholar
- [2] . 2015. A theoretical framework to represent narrative structures for visual storytelling. In Bridges 2015: Highlights of the Conference on Mathematics, Music, Art, Architecture, Culture. 129–136.Google Scholar
- [3] . 2020. On pearl’s hierarchy and the foundations of causal inference. In ACM Special Volume in Honor of Judea Pearl. 507–556.Google Scholar
- [4] . 2016. Can we open the black box of AI? Nature News 538, 7623 (2016), 20–23.Google ScholarCross Ref
- [5] . 2021. Exogenous coordination in multi-scale systems: How information flows and timing affect system properties. Fut. Gener. Comput. Syst. 114 (2021), 403–426.Google ScholarCross Ref
- [6] . 2021. To do or not to do: Finding causal relations in smart homes. In Proceedings of the IEEE International Conference on Autonomic Computing and Self-Organizing Systems (ACSOS’21). IEEE, 110–119.
_eprint: 2105.10058 .Google ScholarCross Ref - [7] . 2014. Challenges of big data analysis. Natl. Sci. Rev. 1, 2 (2014), 293–314.Google ScholarCross Ref
- [8] . 2009. Probabilistic Graphical Models: Principles and Techniques. MIT press.Google Scholar
- [9] . 2016. A simulation-based architecture for smart cyber-physical systems. In Proceedings of the IEEE International Conference on Autonomic Computing (ICAC’16). 374–379. Google ScholarCross Ref
- [10] . 2020. A survey of deep learning techniques for autonomous driving. J. Field Robot. 37, 3 (2020), 362–386.Google ScholarCross Ref
- [11] . 2020. A decentralized approach to explanatory artificial intelligence for autonomic systems. In ACSOS 2020 Conference Proceedings Companion.Google ScholarCross Ref
- [12] . 2010. Causal inference using the algorithmic markov condition. IEEE Trans. Inf. Theory 56, 10 (2010), 5168–5194.Google ScholarDigital Library
- [13] . 1996. An Introduction to Bayesian Networks, Vol. 210. UCL Press, London.Google ScholarDigital Library
- [14] . 2009. Probabilistic Graphical Models: Principles and Techniques. MIT Press.Google ScholarDigital Library
- [15] . 2009. A rigorous architectural approach to adaptive software engineering. J. Comput. Sci. Technol. 24, 2 (2009), 183–188.Google ScholarDigital Library
- [16] . 2015. A survey on engineering approaches for self-adaptive systems. Perv. Mobile Comput. 17 (2015), 184–206.Google ScholarDigital Library
- [17] . 2020. A service-oriented edge platform for cyber-physical systems. CCF Trans. Perv. Comput. Interact. 2, 3 (2020), 206–217.Google ScholarCross Ref
- [18] . 2020. Explainable reinforcement learning through a causal lens. In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 34.
2493–2500 .Google ScholarCross Ref - [19] . 2006. Learning causal bayesian networks from observations and experiments: A decision theoretic approach. In Modeling Decisions for Artificial Intelligence, , , , and (Eds.). Springer Berlin Heidelberg, Berlin, Heidelberg, 58–69.Google Scholar
- [20] . 2018. Learning bayesian network classifiers with completed partially directed acyclic graphs. In International Conference on Probabilistic Graphical Models. PMLR, 272–283.Google Scholar
- [21] . 1986. Fusion, propagation, and structuring in belief networks. Artif. Intell. 29, 3 (1986), 139–188.Google ScholarDigital Library
- [22] . 2009. Causality: Models, Reasoning and Inference (2nd ed.). Cambridge University Press, New York, NY.Google ScholarCross Ref
- [23] . 2018. The Book of Why. Basic Books, New York, NY.Google ScholarDigital Library
- [24] . 2017. Elements of Causal Inference: Foundations and Learning Algorithms. MIT Press.Google ScholarDigital Library
- [25] . 2020. Improving the accuracy of medical diagnosis with causal machine learning. Nat. Commun. 11, 3923 (2020), 1–9.Google Scholar
- [26] . 2017. Mastering the game of go without human knowledge. Nature 550, 7676 (2017), 354–359.Google ScholarCross Ref
- [27] . 1993. Causation, Prediction, and Search. MIT Press.Google ScholarCross Ref
- [28] . 2019. Digital twins and cyber–physical systems toward smart manufacturing and industry 4.0: Correlation and comparison. Engineering 5, 4 (2019), 653–661.Google ScholarCross Ref
- [29] . 2020. Resolving spurious correlations in causal models of environments via interventions. arXiv:2002.05217. Retrieved from https://arxiv.org/abs/2002.05217.Google Scholar
Index Terms
- Improving Causal Learning Scalability and Performance using Aggregates and Interventions
Recommendations
Entropic causal inference
AAAI'17: Proceedings of the Thirty-First AAAI Conference on Artificial IntelligenceWe consider the problem of identifying the causal direction between two discrete random variables using observational data. Unlike previous work, we keep the most general functional model but make an assumption on the unobserved exogenous variable: ...
Learning and testing causal models with interventions
NIPS'18: Proceedings of the 32nd International Conference on Neural Information Processing SystemsWe consider testing and learning problems on causal Bayesian networks as defined by Pearl [Pea09]. Given a causal Bayesian network Μ on a graph with n discrete variables and bounded in-degree and bounded "confounded components", we show that O(log n) ...
Applications of common entropy for causal inference
NIPS '20: Proceedings of the 34th International Conference on Neural Information Processing SystemsWe study the problem of discovering the simplest latent variable that can make two observed discrete variables conditionally independent. The minimum entropy required for such a latent is known as common entropy in information theory. We extend this ...
Comments