Open Problems

Causal Inference & Discovery

Robustness and Adaptation of Causal Frameworks to Arbitrary DAGs and Non-Linear SCMs

UnclassifiedOpen
Strong candidate · 1/1 runs3 papers report this33% from 2025+

Generated automatically from the limitations stated in 3 papers (AISTATS, NeurIPS), listed under Evidence. It is not a paper, and it does not come from papers submitted to CSPaper.

The problem

Current causal frameworks often rely on restrictive scope assumptions: fully known causal graphs restricted to simple canonical topologies (causal, anticausal, FIIF), strict linearity of structural equations, and sensitive feasibility thresholds. In real-world applications, true graphs are rarely known or confined to canonical 3-node topologies, and underlying mechanisms routinely exhibit non-linear dynamics. Consequently, practitioners cannot deploy these methods without risking severe model failure or infeasibility due to unverified structural and functional assumptions.

Why it matters

Enables the principled deployment of causal algorithms in observational settings where domain experts lack complete causal diagrams and linear functional guarantees.

Ways to approach it

Prior-work checks are free with an account. Results someone already ran are shown to everyone.

  1. 1

    Systematic Robustness Benchmarking: Evaluate existing methods across synthetic and semi-synthetic datasets with varying graph topologies (beyond canonical structures to arbitrary DAGs) and functional non-linearities (e.g., post-nonlinear and neural SCMs), measuring performance degradation and feasibility failure rates.

  2. 2

    Partial-Graph Adaptation: Modify existing inference and recourse pipelines to operate over Markov equivalence classes or Partial Ancestral Graphs (PAGs) rather than fully specified graphs, measuring causal estimation accuracy under structural ambiguity.

  3. 3

    Non-Linear Extension via Flexible Function Approximators: Integrate non-linear structural estimators (such as non-linear ICA or normalizing flows) into downstream causal tasks, measuring empirical counterfactual validity and constraint satisfaction against standard linear baselines.

Have a different approach?

Describe how you would tackle this problem and we'll look for papers that already do it. Free; your text stays private.

Free · 3 checks per day

Why it might fail

Fundamental theoretical identifiability limits under observational non-linear SCMs without interventions may prevent point identification, forcing the framework to produce wide, uninformative bounds rather than actionable estimates.

Evidence

Each paper's own statement of the limitation, verbatim.

Nearest existing work

Related open problems

Causal Inference & Discovery

Scope to testPartly addressed

Benchmarking Causal Effect Estimation Across Multi-Valued and Continuous Treatment Regimes

The vast majority of causal inference and discovery algorithms are formulated and evaluated strictly under binary treatment assumptions ($T \in \{0, 1\}$). In real-world domains such as medicine and econometrics, interventions are routinely dose-dependent, multi-categorical, or continuous, rendering binary-only methods inapplicable without ad-hoc binarization. Because existing methods have not been systematically evaluated across non-binary treatment spaces, practitioners lack empirical evidence on how severely estimation and discovery performance degrade when moving beyond the binary regime.

Possible candidate · 3/5 runs10 papers report this60% from 2025+
Generated automatically, not curated by hand. Automated prior-work checks catch about a third of existing work, so treat this problem as a lead to investigate.