The Linux Foundation Projects
Skip to main content

LF AI & Data Foundation—the organization building an ecosystem to sustain open source innovation in artificial intelligence (AI) and data open source projects, today announced DeepCausality as its latest Sandbox Project.

DeepCausality is an advanced hyper-geometric computational causality library tailored for the Rust programming language. The library is engineered to overcome the limitations of conventional deep learning models by focusing on fast and deterministic context-aware causal reasoning. DeepCausality integrates hypergeometric recursive causal models and end-to-end explainability, creating a robust framework for various industries.

Dr. Ibrahim Haddad, Executive Director of LF AI & Data, said: “The addition of DeepCausality to LF AI & Data Foundation further diversifies our growing project portfolio and aligns with our mission to advance and democratize AI and data. DeepCausality, with its focus on computational causality, stands to bring about transformative changes in dynamic systems across industries.”

Key Features of DeepCausality:

  • Hypergeometric recursive causal models: DeepCausality utilizes recursive isomorphic causal data structures to handle structural complexity, allowing causal models to be expressed as graph networks for efficient reasoning over complex structures. These graphs can contain nodes that represent causes, sub-causal graphs, or collections of causes.
  • Contextual causal reasoning: DeepCausality combines causal hypergraphs and contextual hypergraphs to form the backbone of contextual causal reasoning. This approach enables the evaluation of multiple causal models side by side in a shared context, offering flexibility for model generation, evaluation, and re-deployment. With new capabilities added in DC 0.6, causal models can now utilize multiple contexts for even more advanced use cases, streamlining complex scenarios like those in the financial and IoT industries where multiple contextual layers often intersect.
  • End-to-end explainability: DeepCausality offers full explainability through its built-in Graph explanation path. The explanation is constructed based on the path taken through the causal model, providing a clear line of reasoning for each evaluation stage and making it easier to identify any deviations in the model’s behavior.
  • Causal State Machines: DeepCausality introduces a causal state machine that defines states as causes and links them to specific actions. Unlike traditional finite state machines, causal state machines can be dynamically generated at runtime, making it suitable for systems that require dynamic, automated supervision.
  • Context support for fixed and dynamic structures: DeepCausality provides complete support for contextualizing causal models through hypergeometric representation. Context can be either of fixed structure with updated values or of dynamic structure defined at runtime, enabling efficient reasoning over contextualized data.

What Can You Do with DeepCausality?

  • Contextualized Streaming Data: DeepCausality is ideally suited for monitoring solutions using drones, as it can efficiently process multi-dimensional data streams and offer context that conventional deep learning systems might miss.
  • Financial Modeling: DeepCausality addresses the limitations of traditional deep learning in financial markets by providing real-time, context-informed models capable of capturing causal relations across temporal-spatial data.
  • Dynamic Control Systems: The project is particularly beneficial for cloud-native applications requiring dynamic system configurations and monitoring, thanks to its causal state machines.
  • Industries Subject to Safety Regulations: For sectors such as transportation, avionics, or defense, where non-deterministic deep learning is not an option, DeepCausality offers a viable alternative due to its deterministic nature.
  • Combined Deep Causality Learning: Start-ups and enterprises can explore combinations of deep learning and DeepCausality to gain a competitive edge, even in fields like theoretical physics, advanced science, or avionics.

“Joining the LF AI & Data Foundation gives us the platform to collaborate with other leading projects and experts in the field. We are excited to contribute to the community and look forward to the synergies that this collaboration will inevitably create,” states Marvin Hansen, the author of DeepCausality.

Learn more about DeepCausality on their GitHub and join the DeepCausality-Announce Mailing List

A warm welcome to DeepCausality! We are excited to see the project’s continued growth and success as part of the LF AI & Data Foundation. If you are interested in hosting an open source project with us, please visit the LF AI & Data website to learn more.

DeepCausality Key Links

LF AI & Data Resources

Access other resources on LF AI & Data’s GitHub or Wiki