# causality

Causality is the study of causal relationships, including general causal patterns and causation involving particular events.

This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles.

## Causal Discovery with False Positive Error Control

DevFeed: [Causal Discovery with False Positive Error Control](<https://devfeed.tech/articles/causal-discovery-with-false-positive-error-control-145047.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.25376>)

Author: Erik Jahn, Venkat Chandrasekaran, Frederick Eberhardt, Leonard Schulman

Published: 2026-11-20T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [causality](<https://devfeed.tech/topics/causality.md>), [conditionals](<https://devfeed.tech/topics/conditionals.md>)

Tags: [algorithm](<https://devfeed.tech/tags/algorithm.md>), [conditional](<https://devfeed.tech/tags/conditional.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [discovery](<https://devfeed.tech/tags/discovery.md>), [error-control](<https://devfeed.tech/tags/error-control.md>), [faithfulness](<https://devfeed.tech/tags/faithfulness.md>), [false-positive](<https://devfeed.tech/tags/false-positive.md>), [gaussian](<https://devfeed.tech/tags/gaussian.md>), [graph](<https://devfeed.tech/tags/graph.md>), [guarantees](<https://devfeed.tech/tags/guarantees.md>), [sample](<https://devfeed.tech/tags/sample.md>), [stat-ml](<https://devfeed.tech/tags/stat-ml.md>)

### AI overview

The m-conservative PC algorithm controls false discovery rate for causal graphs in the linear Gaussian setting under an extremely weak faithfulness assumption, with guarantees that hold in finite samples despite possible errors in conditional-independence tests. It defines false positives through the conditional-independence information implied by a graph, penalizing both false adjacencies and incorrect edge orientations. The method combines undirected graph learning, a Benjamini-Yekutieli-style conditional-independence testing procedure, and conservative orientation that leaves unresolved or contradictory directions ambiguous. Its parameter m trades algorithmic complexity against the strength of the required faithfulness assumption.

### Source excerpt

arXiv:2611.25376v1 Announce Type: cross Abstract: Causal discovery methods produce data-driven hypotheses about causal relationships, which may become new scientific findings or inform decisions about downstream interventions. In such settings, false positive causal claims can be more consequential than false negatives. Yet few methods target false positive error control in causal discovery, and existing error guarantees are often asymptotic and require strong versions of the faithfulness assumption. We introduce the $m$-conservative PC algorithm and prove that it controls false discovery rate in the linear Gaussian setting under an extremely weak version of the faithfulness assumption. In our framework, false positive error for causal graphs is defined based on the conditional-independence information they entail, yielding a metric that penalizes both false adjacencies and incorrect edge orientations. The parameter $m$ quantifies a trade-off between algorithmic complexity and the strength of the required faithfulness assumption. The $m$-conservative PC algorithm combines undirected graph learning, a Benjamini-Yekutieli-style procedure for conditional independence testing, and a conservative orientation step that marks unresolved or contradictory edge directions as ambiguous. Our algorithm produces valid outputs and retains error guarantees even in the finite-sample setting, despite possible random errors in the conditional independence tests.

## Causal Overtaking in Fair Transaction Ordering

DevFeed: [Causal Overtaking in Fair Transaction Ordering](<https://devfeed.tech/articles/causal-overtaking-in-fair-transaction-ordering-146054.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.26100>)

Author: Haozhe Feng, Di Wu, Jian Liu

Published: 2026-11-20T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Cryptography and Security](<https://devfeed.tech/sources/arxiv-cryptography-and-security.md>)

Topics: [ordering](<https://devfeed.tech/topics/ordering.md>), [Simulation](<https://devfeed.tech/topics/simulation.md>), [causality](<https://devfeed.tech/topics/causality.md>), [Latency](<https://devfeed.tech/topics/latency.md>)

Tags: [attack](<https://devfeed.tech/tags/attack.md>), [aws](<https://devfeed.tech/tags/aws.md>), [cs-cr](<https://devfeed.tech/tags/cs-cr.md>), [latency](<https://devfeed.tech/tags/latency.md>), [measurements](<https://devfeed.tech/tags/measurements.md>), [mitigation](<https://devfeed.tech/tags/mitigation.md>), [model](<https://devfeed.tech/tags/model.md>), [network](<https://devfeed.tech/tags/network.md>), [ordering](<https://devfeed.tech/tags/ordering.md>), [protocols](<https://devfeed.tech/tags/protocols.md>), [simulations](<https://devfeed.tech/tags/simulations.md>), [transaction](<https://devfeed.tech/tags/transaction.md>)

### AI overview

The paper identifies causal overtaking in fair transaction ordering: corrupted replicas can submit identical signed copies of a responsive transaction through multiple client entries, letting their first arrivals collectively provide ordering support that no single path achieves. The authors formalize receipt-order protocols including Themis, characterize when the attack is feasible, and report that delayed gossip does not prevent it. Under their network model, opposite required transaction orders can yield identical receipt evidence, so receipt-only mechanisms cannot always preserve earlier-submission priority while preventing responsive overtaking. They propose a mitigation and evaluate the attack and mitigation on Themis using simulations driven by AWS and WonderProxy latency measurements; the abstract does not describe the mitigation's design.

### Source excerpt

arXiv:2611.26100v1 Announce Type: new Abstract: Fair transaction ordering protocols aim to mitigate front-running using replica-local arrival orders, but prior work often assumes a single submission origin for each responsive transaction. We identify causal overtaking, in which corrupted replicas observe a target and submit identical signed copies of a responsive transaction through multiple client entries. The copies' first arrivals at different replicas can collectively provide ordering support that no individual path achieves, allowing the response to execute before its target. We formalize a class of receipt-order protocols, including Themis, characterize the timing and network conditions under which the attack is feasible, and show that delayed gossip does not rule it out. We further prove that, under our network model, histories requiring opposite transaction orders can produce identical receipt evidence. Thus, receipt-only mechanisms cannot always preserve earlier-submission priority while preventing responsive overtaking. We propose a mitigation and analyze its assumptions and limitations. Finally, we evaluate the attack and mitigation on Themis through simulations driven by AWS and WonderProxy latency measurements.

## Time Series Structure Learning with Local Causal Parent Discovery

DevFeed: [Time Series Structure Learning with Local Causal Parent Discovery](<https://devfeed.tech/articles/time-series-structure-learning-with-local-causal-parent-discovery-145154.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.26232>)

Author: Jacqueline Maasch, Shu Wan, Hank Zhu, Huan Liu, Zhiguang Wang

Published: 2026-11-20T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Time Series](<https://devfeed.tech/topics/time-series.md>), [causality](<https://devfeed.tech/topics/causality.md>)

Tags: [algorithms](<https://devfeed.tech/tags/algorithms.md>), [complexity](<https://devfeed.tech/tags/complexity.md>), [condition](<https://devfeed.tech/tags/condition.md>), [constraint](<https://devfeed.tech/tags/constraint.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [discovery](<https://devfeed.tech/tags/discovery.md>), [distribution](<https://devfeed.tech/tags/distribution.md>), [exploits](<https://devfeed.tech/tags/exploits.md>), [f1](<https://devfeed.tech/tags/f1.md>), [faster](<https://devfeed.tech/tags/faster.md>), [frontier](<https://devfeed.tech/tags/frontier.md>), [global](<https://devfeed.tech/tags/global.md>), [graph](<https://devfeed.tech/tags/graph.md>), [graphs](<https://devfeed.tech/tags/graphs.md>), [learning](<https://devfeed.tech/tags/learning.md>), [markov](<https://devfeed.tech/tags/markov.md>), [nodes](<https://devfeed.tech/tags/nodes.md>), [oracle](<https://devfeed.tech/tags/oracle.md>), [series](<https://devfeed.tech/tags/series.md>), [stat-ml](<https://devfeed.tech/tags/stat-ml.md>), [structure](<https://devfeed.tech/tags/structure.md>), [time-series](<https://devfeed.tech/tags/time-series.md>)

### AI overview

Iterative Temporal Parent Discovery (ITPD) learns causal structure in multivariate time series with causal nonstationarity by discovering each variable's parent set in sequence. It uses temporal structural priors, graph partitioning, and the local Markov condition to replace higher-order conditional independence tests with lower-order ones. With an exact oracle, it used the fewest tests and smallest average conditioning sets among the compared baselines. In simulations, it lay on the F1 and runtime Pareto frontier and was faster than baselines on graphs with at least 1,000 nodes; on 2,000-node graphs, it ran 2.3x faster than iterative IAMB, 9.1x faster than DF-PC, and 40.9x faster than PCMCI.

### Source excerpt

arXiv:2611.26232v1 Announce Type: new Abstract: Learning and inference on time series can pose significant statistical challenges, including autocorrelated observations, distribution shift over time, and the curse of dimensionality. These challenges complicate the practical application of causal discovery algorithms, which already suffer from high computational and sample complexity in the general case. To mitigate these issues, this work introduces Iterative Temporal Parent Discovery (ITPD): a sound and complete constraint-based procedure for causal structure learning on multivariate time series with causal nonstationarity. ITPD takes a local-to-global learning approach that discovers the global graph by sequentially discovering the parent set for each variable. ITPD exploits temporal structural priors, graph partitioning, and the local Markov condition to swap higher-order conditional independence tests for lower-order alternatives. With an exact oracle, ITPD performed the least number of tests and used the smallest average conditioning set sizes relative to baselines. In benchmarking simulations, ITPD sat on the Pareto frontier with respect to F1 and wall-clock time and was faster than baselines on large graphs ($\geq$ 1000 nodes). For graphs with 2000 nodes, ITPD ran $2.3\times$, $9.1\times$, and $40.9\times$ faster than iterative IAMB, DF-PC, and PCMCI, respectively.

## Evidence from a Multi-Regime Causal Discovery Benchmark: UCMT Algorithms, Comparative Performance, and Controlled Ablation Studies

DevFeed: [Evidence from a Multi-Regime Causal Discovery Benchmark: UCMT Algorithms, Comparative Performance, and Controlled Ablation Studies](<https://devfeed.tech/articles/evidence-from-a-multi-regime-causal-discovery-benchmark-ucmt-algorithms-comparative-performance-and-controlled-ablation-studies-145172.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.26383>)

Author: Mahyar Eslami

Published: 2026-11-20T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [causality](<https://devfeed.tech/topics/causality.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [reproducibility](<https://devfeed.tech/topics/reproducibility.md>)

Tags: [ablation](<https://devfeed.tech/tags/ablation.md>), [algorithms](<https://devfeed.tech/tags/algorithms.md>), [benchmark](<https://devfeed.tech/tags/benchmark.md>), [comparison](<https://devfeed.tech/tags/comparison.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [data](<https://devfeed.tech/tags/data.md>), [discovery](<https://devfeed.tech/tags/discovery.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [f1](<https://devfeed.tech/tags/f1.md>), [multiple-testing](<https://devfeed.tech/tags/multiple-testing.md>), [orientation](<https://devfeed.tech/tags/orientation.md>), [performance](<https://devfeed.tech/tags/performance.md>), [permutation](<https://devfeed.tech/tags/permutation.md>)

### AI overview

In this internal, pre-registered benchmark, UCMT v2 Global beats NOTEARS and DAGMA on directed F1 in selected comparisons but trails leading permutation-search methods in other settings. UCMT v4 improves over BOSS on CSuite at the paired-run level (+0.279; Holm p < 0.001), though the dataset-level adjusted result is not significant (p = 0.188). In a separate fixed-skeleton study on standardized iSCM data, v2 Global exceeds an adapted R²-SortnRegress orientation baseline (0.537 vs 0.468; dataset-level Holm p = 0.031), with absolute orientation accuracy of 0.563 and no established broad advantage over RESIT, ANM, or DirectLiNGAM. Ablations indicate sparsity and BIC-based refinement affect some outcomes; arbitration and automatic regime selection show no confirmed general improvement. The evidence supports empirical investigation of the UCMT family, not overall superiority or independent validation.

### Source excerpt

arXiv:2611.26383v1 Announce Type: new Abstract: We present an internal, pre-registered evaluation of the UCMT causal discovery algorithm family alongside established alternatives across heterogeneous benchmark regimes, complemented by a separately pre-registered, controlled study of UCMT v2 Global components. The contribution is both empirical and methodological: a family-stratified comparison of three actively developed algorithm configurations and targeted evidence about which internal stages affect their outcomes. In the main benchmark, UCMT v2 Global has higher directed F1 than NOTEARS and DAGMA in selected real, semi-synthetic and Gaussian-network comparisons at the paired-run level, while trailing leading permutation-search methods in other settings. The BOSS-based UCMT v4 produces a CSuite-specific directed-F1 improvement over BOSS (mean paired difference +0.279; pair-level Holm p < 0.001), but the dataset-level adjusted result is not significant (p = 0.188). In a separate fixed-skeleton ablation, UCMT v2 Global exceeds an adapted R^2-SortnRegress orientation baseline on standardized iSCM data (directed F1 0.537 vs 0.468, dataset-level Holm p = 0.031); absolute orientation accuracy is 0.563 and there is no established broad advantage over RESIT, ANM or DirectLiNGAM. Controlled ablations show that sparsity and BIC-based refinement materially affect some outcomes, whereas the examined arbitration and automatic regime-selection mechanisms do not yield confirmed general improvements. We report performance shortcomings, multiple-testing correction, runtime costs, evaluation limitations and reproducibility checks. The results support UCMT as an implemented and empirically investigated algorithm family, not a claim of overall superiority, independent validation or demonstrated commercial impact.

## Towards Generalizable Reasoning: Group Causal Counterfactual Policy Optimization for LLM Reasoning

DevFeed: [Towards Generalizable Reasoning: Group Causal Counterfactual Policy Optimization for LLM Reasoning](<https://devfeed.tech/articles/towards-generalizable-reasoning-group-causal-counterfactual-policy-optimization-for-llm-reasoning-143232.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2602.06475>)

Author: Jingyao Wang, Xingyu Zhang, Peizheng Guo, Jiahuan Zhou, Zhenghao Zhang, Changwen Zheng, Chuxiong Sun, Wenwen Qiang

Published: 2026-11-19T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [llm-reasoning](<https://devfeed.tech/topics/llm-reasoning.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [causality](<https://devfeed.tech/topics/causality.md>)

Tags: [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [large-language-models-llms](<https://devfeed.tech/tags/large-language-models-llms.md>), [llm-reasoning](<https://devfeed.tech/tags/llm-reasoning.md>), [llms](<https://devfeed.tech/tags/llms.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [optimization](<https://devfeed.tech/tags/optimization.md>), [reasoning](<https://devfeed.tech/tags/reasoning.md>)

### AI overview

Group Causal Counterfactual Policy Optimization aims to improve LLM reasoning generalization by rewarding reasoning processes, not just final-answer correctness. The authors frame multiple candidate reasoning paths for a question as counterfactual experiments. Their episodic reward combines robustness--keeping the answer distribution induced by a reasoning step stable under counterfactual perturbations--with effectiveness, requiring enough variability for a strategy to transfer across questions. They derive token-level advantages from this reward to optimize the policy toward process-valid, counterfactually robust reasoning. The abstract reports advantages on diverse benchmarks but provides no benchmark details or results.

### Source excerpt

arXiv:2602.06475v3 Announce Type: replace Abstract: Large language models (LLMs) excel at complex tasks with advances in reasoning capabilities. However, existing reward mechanisms remain tightly coupled to final correctness and pay little attention to the underlying reasoning process: trajectories with sound reasoning but wrong answers receive low credit, while lucky guesses with flawed logic may be highly rewarded, affecting reasoning generalization. From a causal perspective, we interpret multi-candidate reasoning for a fixed question as a family of counterfactual experiments. Building on this, we propose Group Causal Counterfactual Policy Optimization to explicitly train LLMs to learn generalizable reasoning patterns. It proposes an episodic causal counterfactual reward that jointly captures (i) robustness, encouraging the answer distribution induced by a reasoning step to remain stable under counterfactual perturbations; and (ii) effectiveness, enforcing sufficient variability so that the learned reasoning strategy can transfer across questions. We then construct token-level advantages from this reward and optimize the policy, encouraging LLMs to favor reasoning patterns that are process-valid and counterfactually robust. Extensive experiments on diverse benchmarks demonstrate its advantages.

## Fill-Distance-Guided Active Learning for Causal Function Recovery under Network Interference

DevFeed: [Fill-Distance-Guided Active Learning for Causal Function Recovery under Network Interference](<https://devfeed.tech/articles/fill-distance-guided-active-learning-for-causal-function-recovery-under-network-interference-143487.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.24143>)

Author: Peiyao Liu, Hongtao Zhu, Sizhe Zhang, Yang Su, Zhenyu Zhao, Nan Chen

Published: 2026-11-19T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [causality](<https://devfeed.tech/topics/causality.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>)

Tags: [budget](<https://devfeed.tech/tags/budget.md>), [causal-inference](<https://devfeed.tech/tags/causal-inference.md>), [conditional](<https://devfeed.tech/tags/conditional.md>), [coverage](<https://devfeed.tech/tags/coverage.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [design](<https://devfeed.tech/tags/design.md>), [effects](<https://devfeed.tech/tags/effects.md>), [error](<https://devfeed.tech/tags/error.md>), [experiment](<https://devfeed.tech/tags/experiment.md>), [experimental-design](<https://devfeed.tech/tags/experimental-design.md>), [experimentation](<https://devfeed.tech/tags/experimentation.md>), [full](<https://devfeed.tech/tags/full.md>), [function](<https://devfeed.tech/tags/function.md>), [gaussian](<https://devfeed.tech/tags/gaussian.md>), [global](<https://devfeed.tech/tags/global.md>), [inference](<https://devfeed.tech/tags/inference.md>), [learning](<https://devfeed.tech/tags/learning.md>), [model](<https://devfeed.tech/tags/model.md>), [network](<https://devfeed.tech/tags/network.md>), [online](<https://devfeed.tech/tags/online.md>), [recovery](<https://devfeed.tech/tags/recovery.md>), [stat-me](<https://devfeed.tech/tags/stat-me.md>)

### AI overview

FANCI uses sequential experiments to recover a population-average causal response function when outcomes depend on both a unit's treatment and its neighbors' treatments. Treatment-specific Gaussian processes model nonlinear response surfaces, while a fill-distance criterion selects experiments in poorly covered exposure regions. The method offers full-network and target-screened strategies for choosing feasible network-assignment pairs; the abstract says it derives uniform error bounds connecting fill distance to estimation error and characterizes the approximation loss from target screening. Simulations and an online-game experiment reportedly show improved recovery of nonlinear effects and treatment patterns obscured by ignoring interference. The supplied text is an abstract, so it does not provide implementation details or quantitative results.

### Source excerpt

arXiv:2611.24143v1 Announce Type: cross Abstract: Causal inference seeks to learn how outcomes change under alternative interventions, but network interactions complicate this task. Under network interference, outcomes depend on both own treatment and neighbors' treatments, making causal effects functions of neighborhood exposure rather than single treatment contrasts. Existing methods primarily target prespecified causal contrasts or their estimation precision, leaving global response-function recovery largely unexplored. We study sequential experimentation under a limited experimental budget, with the goal of recovering the full exposure-dependent population-average response function. To this end, we propose FANCI (Fill-Distance-Guided Active Learning for Network-Interference Causal Inference). For function learning, treatment-specific Gaussian processes flexibly model nonlinear conditional response surfaces. For experimental design, we formulate global response-function recovery as a coverage problem and propose a fill-distance criterion that guides new experiments toward poorly covered regions. To account for network constraints on attainable exposure patterns, we develop full-network and target-screened strategies for selecting feasible network--assignment pairs. Theoretically, we derive uniform error bounds linking fill distance to response-function estimation and characterize the approximation loss from target screening. Simulations and an online-game experiment show that FANCI improves recovery of nonlinear exposure-dependent effects and reveals treatment patterns obscured when network interference is ignored.

## Attention Is Not Relevance: Challenging the Causality of Attention in Autonomous Driving

DevFeed: [Attention Is Not Relevance: Challenging the Causality of Attention in Autonomous Driving](<https://devfeed.tech/articles/attention-is-not-relevance-challenging-the-causality-of-attention-in-autonomous-driving-142746.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.24051>)

Author: Luca Lusvarghi, Javier Gozalvez

Published: 2026-11-19T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [autonomous vehicles](<https://devfeed.tech/topics/autonomous-vehicles.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [causality](<https://devfeed.tech/topics/causality.md>), [Transformer](<https://devfeed.tech/topics/transformer.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [attention](<https://devfeed.tech/tags/attention.md>), [autonomous-driving](<https://devfeed.tech/tags/autonomous-driving.md>), [autonomous-vehicle](<https://devfeed.tech/tags/autonomous-vehicle.md>), [causality](<https://devfeed.tech/tags/causality.md>), [counterfactual-analysis](<https://devfeed.tech/tags/counterfactual-analysis.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [natural-language-processing](<https://devfeed.tech/tags/natural-language-processing.md>), [planning](<https://devfeed.tech/tags/planning.md>), [relevance](<https://devfeed.tech/tags/relevance.md>), [synthetic](<https://devfeed.tech/tags/synthetic.md>), [transformer](<https://devfeed.tech/tags/transformer.md>)

### AI overview

Across more than 80,000 synthetic driving scenes evaluated with a vehicle based on PlanT 2.0, attention scores were correlational rather than reliable measures of how much individual objects causally affect driving decisions. Compared with counterfactual estimates of object relevance, attention methods missed some relevant objects and often labeled objects with little or no impact as relevant. They performed effectively in simple, straight single-lane scenarios, but accuracy declined on multi-lane roads and at intersections. The findings caution against using attention alone as a proxy for causal object relevance; the abstract calls for methods that distinguish causal impact from correlation.

### Source excerpt

arXiv:2611.24051v1 Announce Type: new Abstract: Attention has become one of the most widely adopted solutions for estimating the relevance of surrounding objects and interpreting an autonomous vehicle's driving decisions. However, its ability to capture the causal impact of individual input elements on an AI model's output has long been discussed in the literature, particularly in the Natural Language Processing (NLP) domain. This work brings this discussion into the autonomous driving domain by analyzing the accuracy of attention in estimating the impact of an input object on an autonomous vehicle's driving, i.e., its relevance. To this end, we compare different attention-based relevance estimation methods against a causal reference of object relevance obtained through counterfactual analysis. We perform our analysis on more than 80,000 synthetic driving scenes using an autonomous vehicle based on PlanT 2.0, a state-of-the-art Transformer-based model for object-level planning. Results show that attention provides a correlational rather than causal indication of object relevance. Attention-based relevance estimation methods do not consistently identify all relevant objects and tend to incorrectly classify as relevant objects with little or no impact on the vehicle's driving. In particular, results show that attention can effectively identify relevant objects in simple single-lane, straight-road scenarios, but its accuracy degrades in more complex settings, such as multi-lane roads and intersections. These findings highlight the limitations of attention as a proxy for object relevance, calling for novel methods able to accurately capture the causal impact of individual objects on the vehicle's driving and distinguish it from correlational associations.

## Graph-of-Thoughts for LLM Causal Discovery with a Fairness-Pathway Case Study

DevFeed: [Graph-of-Thoughts for LLM Causal Discovery with a Fairness-Pathway Case Study](<https://devfeed.tech/articles/graph-of-thoughts-for-llm-causal-discovery-with-a-fairness-pathway-case-study-143295.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2608.02877>)

Author: Nitish Nagesh, Elahe Khatibi, Thomas Dean Hughes, Mahdi Bagheri, Pratik Gajane, Amir M. Rahmani

Published: 2026-11-19T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [llm-reasoning](<https://devfeed.tech/topics/llm-reasoning.md>), [causality](<https://devfeed.tech/topics/causality.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [responsible-ai](<https://devfeed.tech/topics/responsible-ai.md>)

Tags: [case-study](<https://devfeed.tech/tags/case-study.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [discovery](<https://devfeed.tech/tags/discovery.md>), [fairness](<https://devfeed.tech/tags/fairness.md>), [graph](<https://devfeed.tech/tags/graph.md>), [large-language-models-llms](<https://devfeed.tech/tags/large-language-models-llms.md>), [llm](<https://devfeed.tech/tags/llm.md>)

### AI overview

GoT-CD generates several complete causal graph candidates from variable metadata, ranks them with a deterministic validity score, merges and refines them, then projects the result to a DAG. In tests on Asia, Child, and an Alzheimer's graph with 100 samples using GPT-4o-mini, it achieved the strongest mean F1 and lowest mean structural Hamming distance among the LLM methods on Child and Alzheimer's across five seeds, returned a DAG in every run, and used three LLM requests per graph. Ablations attributed most of the gain to generating and merging multiple candidates. Yet structural scores did not ensure a reliable downstream fairness assessment: the discovered graph's verdict about Sex and MOCA Score disagreed with the reference in 12 of 15 LLM runs, including all five GoT-CD runs, which mostly omitted the edge out of Sex. The authors recommend checking pathways relevant to the intended analysis.

### Source excerpt

arXiv:2608.02877v2 Announce Type: replace Abstract: Methods that use large language models (LLMs) for causal discovery reason locally, deciding one variable pair or one node at a time, so they never compare alternative whole graphs. We introduce GoT-CD, a Graph-of-Thoughts method that asks an LLM for several complete candidate graphs from variable metadata alone, ranks them with a deterministic validity score, merges them under a hard union constraint, refines the merge, and projects the result to a DAG. We evaluate on Asia, Child, and an Alzheimer's disease graph with 100 samples and GPT-4o-mini. Over five seeds on Alzheimer's and Child, GoT-CD gives the highest mean F1 score and the lowest mean structural Hamming distance of the LLM methods, with 95% intervals of the difference excluding zero in seven of eight comparisons, and it returns a DAG in every run while using three LLM requests per graph. Replaying the logged runs without individual components attributes most of the gain to generating and merging several complete candidates. A case study on the effect of Sex on MOCA Score in the Alzheimer's graph shows the limits of structural scores: the fairness verdict drawn from the discovered graph disagrees with the reference verdict in twelve of fifteen LLM runs, including all five GoT-CD runs, which mostly drop the edge out of Sex. Graphs meant for a specific downstream analysis should therefore also be checked on the pathways that analysis depends on.

## NOTREKS: Marginal Independence in Score-based Structure Learning

DevFeed: [NOTREKS: Marginal Independence in Score-based Structure Learning](<https://devfeed.tech/articles/notreks-marginal-independence-in-score-based-structure-learning-143471.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.24015>)

Author: Fabian Bleile, Sarah Lumpp, Mathias Drton

Published: 2026-11-19T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [causality](<https://devfeed.tech/topics/causality.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Graphs](<https://devfeed.tech/topics/graphs.md>)

Tags: [adjustment](<https://devfeed.tech/tags/adjustment.md>), [constraint](<https://devfeed.tech/tags/constraint.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [data](<https://devfeed.tech/tags/data.md>), [discovery](<https://devfeed.tech/tags/discovery.md>), [effect](<https://devfeed.tech/tags/effect.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [gradient](<https://devfeed.tech/tags/gradient.md>), [graphs](<https://devfeed.tech/tags/graphs.md>), [independent](<https://devfeed.tech/tags/independent.md>), [information](<https://devfeed.tech/tags/information.md>), [learning](<https://devfeed.tech/tags/learning.md>), [models](<https://devfeed.tech/tags/models.md>), [optimization](<https://devfeed.tech/tags/optimization.md>), [optimizing](<https://devfeed.tech/tags/optimizing.md>), [oracle](<https://devfeed.tech/tags/oracle.md>), [order](<https://devfeed.tech/tags/order.md>), [pairs](<https://devfeed.tech/tags/pairs.md>), [power](<https://devfeed.tech/tags/power.md>), [score](<https://devfeed.tech/tags/score.md>), [stat-me](<https://devfeed.tech/tags/stat-me.md>), [stat-ml](<https://devfeed.tech/tags/stat-ml.md>), [structure](<https://devfeed.tech/tags/structure.md>)

### AI overview

Under faithfulness, known marginal independence between selected variable pairs implies a no-trek constraint on the causal graph. The paper calls this NOTREKS information and incorporates it into score-based structure learning in two ways: a power-series trek (PSTrek) penalty for gradient-based optimization and a trek constraint for order-based search. The authors analyze how the information can help in finite samples and restrict the search space. Synthetic benchmarks using oracle NOTREKS information show improved structural recovery and adjustment-based effect identification; the supplied abstract does not quantify the gains.

### Source excerpt

arXiv:2611.24015v1 Announce Type: cross Abstract: Score-based structure learning estimates causal graphs by optimizing a data-dependent objective over structural equation models. We study how prior knowledge that selected pairs of variables are marginally independent can be incorporated into such causal discovery methods. Under faithfulness, marginally independent pairs cannot be connected by a trek, so the knowledge becomes a graphical no-trek constraint, which we call NOTREKS information. We introduce a power-series trek (PSTrek) penalty for gradient-based optimization and a trek constraint for order-based search, and we analyze why such information helps in finite samples and how it restricts the search space. Our numerical experiments illustrate to what extent NOTREKS information helps in causal discovery. In particular, using synthetic benchmarks with oracle NOTREKS information, we show that these constraints can improve both structural recovery and adjustment-based effect identification.

## Auditing Path-Specific Fairness Across Causal Discovery Algorithms: A Heart Failure Case Study

DevFeed: [Auditing Path-Specific Fairness Across Causal Discovery Algorithms: A Heart Failure Case Study](<https://devfeed.tech/articles/auditing-path-specific-fairness-across-causal-discovery-algorithms-a-heart-failure-case-study-142512.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2603.15926>)

Author: Nitish Nagesh, Elahe Khatibi, Thomas Hughes, Mahdi Bagheri, Pratik Gajane, Amir M. Rahmani

Published: 2026-11-19T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [responsible-ai](<https://devfeed.tech/topics/responsible-ai.md>), [causality](<https://devfeed.tech/topics/causality.md>), [Graph Mining](<https://devfeed.tech/topics/graph-mining.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [algorithm](<https://devfeed.tech/tags/algorithm.md>), [algorithms](<https://devfeed.tech/tags/algorithms.md>), [allows](<https://devfeed.tech/tags/allows.md>), [analysis](<https://devfeed.tech/tags/analysis.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [assessment](<https://devfeed.tech/tags/assessment.md>), [auditing](<https://devfeed.tech/tags/auditing.md>), [benchmark](<https://devfeed.tech/tags/benchmark.md>), [bias](<https://devfeed.tech/tags/bias.md>), [can](<https://devfeed.tech/tags/can.md>), [case](<https://devfeed.tech/tags/case.md>), [case-study](<https://devfeed.tech/tags/case-study.md>), [choice](<https://devfeed.tech/tags/choice.md>), [contribution](<https://devfeed.tech/tags/contribution.md>), [cost](<https://devfeed.tech/tags/cost.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [discovery](<https://devfeed.tech/tags/discovery.md>), [failure](<https://devfeed.tech/tags/failure.md>), [fairness](<https://devfeed.tech/tags/fairness.md>), [study](<https://devfeed.tech/tags/study.md>)

### AI overview

The study finds that causal discovery algorithm choice can change path-specific fairness estimates and the fairness-utility trade-off for heart failure risk models. Using an expert-elicited graph benchmark, it reports that ICA-LiNGAM had the lowest structural Hamming distance and highest recall, while GES had the highest precision. The mediated effect on ejection fraction differed by 1.2 percentage points between the original and discovered graphs, contributing to variation in the fairness-utility ratio. The authors argue these findings support fine-grained, discovery-aware fairness audits; the supplied abstract does not detail the methods or broader limitations.

### Source excerpt

arXiv:2603.15926v2 Announce Type: replace-cross Abstract: Causal discovery is increasingly applied to healthcare data, where the recovered relationships can encode biased pathways that propagate to downstream prediction. Path-specific fairness analysis, through direct, indirect, and spurious effect decompositions along with fairness-utility ratio assessment allows bias quantification and associated prediction cost. However, existing evaluations of causal discovery under fairness constraints have largely used non-clinical datasets and have not separated the contribution of confounders from that of mediators, leaving open how discovery-algorithm choice propagates through path-specific fairness to the utility trade-off in clinical settings. This work addresses that gap by establishing an expert-elicited causal graph benchmark for the heart failure clinical records dataset and tracing discovery algorithms through composite, per-variable fairness decompositions and fairness-utility ratio. ICA-LiNGAM achieves the lowest structural Hamming distance and highest recall, while GES achieves the highest precision. For path-specific effects, there was a 1.2 percentage-point difference in the mediated effect on ejection fraction on overall causal fairness between the original and discovered graphs. This ultimately led to variations in the fairness-utility ratio highlighting need for fine-grained evaluation. These results indicate that discovery-aware, path-specific fairness evaluation is consequential in clinical settings and warrants inclusion in routine auditing of clinical risk models.

## Balancing Partition Forest for Causal Weighting

DevFeed: [Balancing Partition Forest for Causal Weighting](<https://devfeed.tech/articles/balancing-partition-forest-for-causal-weighting-143403.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.23524>)

Author: Felipe Angelim, Alessandro Leite

Published: 2026-11-19T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [causality](<https://devfeed.tech/topics/causality.md>)

Tags: [bias](<https://devfeed.tech/tags/bias.md>), [causal-inference](<https://devfeed.tech/tags/causal-inference.md>), [continuous](<https://devfeed.tech/tags/continuous.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [independent](<https://devfeed.tech/tags/independent.md>), [information](<https://devfeed.tech/tags/information.md>), [rank](<https://devfeed.tech/tags/rank.md>), [ratio](<https://devfeed.tech/tags/ratio.md>), [reduce](<https://devfeed.tech/tags/reduce.md>), [residual](<https://devfeed.tech/tags/residual.md>), [response](<https://devfeed.tech/tags/response.md>), [robust](<https://devfeed.tech/tags/robust.md>), [semi](<https://devfeed.tech/tags/semi.md>), [settings](<https://devfeed.tech/tags/settings.md>), [split](<https://devfeed.tech/tags/split.md>), [stat-me](<https://devfeed.tech/tags/stat-me.md>), [statistics](<https://devfeed.tech/tags/statistics.md>), [synthetic](<https://devfeed.tech/tags/synthetic.md>), [weight](<https://devfeed.tech/tags/weight.md>)

### AI overview

Balancing Partition Tree (BPT) and Balancing Partition Forest (BPF) estimate causal balancing weights by partitioning covariate-treatment space into learned rectangles. At each split, they maximize a Bregman-divergence gain based on joint and marginal counts; alternative divergences target mutual information captured by the partition or lower squared weight error. The authors show that each criterion's population residual bounds bias for inverse-probability-weighted and doubly robust response estimators using population partition weights. Across 17 synthetic and semi-synthetic settings and three response estimators, BPF had the lowest mean RMSE rank among seven weight estimators. A single tree fit in 0.14 seconds on 16,512 observations.

### Source excerpt

arXiv:2611.23524v1 Announce Type: cross Abstract: Balancing weights are used in observational causal inference to construct a population in which the treatment is independent of observed pretreatment covariates. The ideal weight is the ratio between the product of the covariate and treatment marginals and their joint distribution, for binary and continuous treatments alike. We propose Balancing Partition Tree (BPT) and Balancing Partition Forest (BPF), which estimate this ratio on a learned rectangular partition of covariate--treatment space. Each split maximizes the gain in a Bregman divergence computed from joint and marginal counts. Two choices of divergence yield criteria that maximize the mutual information captured by the partition or reduce the squared error of the weights. We show that the population residual of either criterion bounds the bias of inverse-probability-weighted and doubly robust response estimators that use the population partition weights. Across 17 synthetic and semi-synthetic settings and three response estimators, BPF obtains the lowest mean RMSE rank among seven weight estimators, and a single tree fits in 0.14 seconds on 16,512 observations.

## Simpson's Paradox: Beware the Lurking Variable | Blog | Fivetran

DevFeed: [Simpson's Paradox: Beware the Lurking Variable | Blog | Fivetran](<https://devfeed.tech/articles/simpson-s-paradox-beware-the-lurking-variable-blog-fivetran-140254.md>)

Original publisher: [Read original article](<https://wf.fivetran.com/blog/simpsons-paradox>)

Author: Charles Wang

Published: 2026-11-18T05:58:59Z

Content type: article

Language: en

Sources: [Fivetran Blog](<https://devfeed.tech/sources/fivetran-blog.md>)

Topics: [causality](<https://devfeed.tech/topics/causality.md>), [Data Mining & Modeling](<https://devfeed.tech/topics/data-mining-modeling.md>)

Tags: [berkeley](<https://devfeed.tech/tags/berkeley.md>), [blog](<https://devfeed.tech/tags/blog.md>), [business](<https://devfeed.tech/tags/business.md>), [causality](<https://devfeed.tech/tags/causality.md>), [civil-rights](<https://devfeed.tech/tags/civil-rights.md>), [data](<https://devfeed.tech/tags/data.md>), [democratic](<https://devfeed.tech/tags/democratic.md>), [democrats](<https://devfeed.tech/tags/democrats.md>), [department](<https://devfeed.tech/tags/department.md>), [dimensions](<https://devfeed.tech/tags/dimensions.md>), [discrimination](<https://devfeed.tech/tags/discrimination.md>), [earnings](<https://devfeed.tech/tags/earnings.md>), [economy](<https://devfeed.tech/tags/economy.md>), [education](<https://devfeed.tech/tags/education.md>), [election](<https://devfeed.tech/tags/election.md>), [examples](<https://devfeed.tech/tags/examples.md>), [like](<https://devfeed.tech/tags/like.md>), [slice](<https://devfeed.tech/tags/slice.md>), [variable](<https://devfeed.tech/tags/variable.md>), [variables](<https://devfeed.tech/tags/variables.md>)

### AI overview

Simpson's Paradox can make an overall comparison reverse or obscure the patterns within subgroups when group proportions differ. The article illustrates this with Civil Rights Act votes split by region, Berkeley admissions split by department, election preferences linked to race, kidney stone outcomes linked to case severity, and earnings trends affected by changing education levels. For business analysis, it recommends checking relevant dimensions and outside influences, examining data beyond aggregate results, and building a coherent account of how the measurements arose. The examples show why association in aggregated figures alone can mislead; the article does not offer a formal method for establishing causality.

### Source excerpt

You can easily reach misleading conclusions if you don't carefully slice your data, consider lurking variables, and think about causality.

## Scaling Causal Reasoning with Increasingly Complex Causal Simulators

DevFeed: [Scaling Causal Reasoning with Increasingly Complex Causal Simulators](<https://devfeed.tech/articles/scaling-causal-reasoning-with-increasingly-complex-causal-simulators-139766.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2605.09079>)

Author: Nicol\\'as Astorga, Anita Kriz, Mihaela van der Schaar

Published: 2026-11-18T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [causality](<https://devfeed.tech/topics/causality.md>), [llm-reasoning](<https://devfeed.tech/topics/llm-reasoning.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>)

Tags: [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [data-augmentation](<https://devfeed.tech/tags/data-augmentation.md>), [data-volume](<https://devfeed.tech/tags/data-volume.md>), [domain](<https://devfeed.tech/tags/domain.md>), [large-language-models-llms](<https://devfeed.tech/tags/large-language-models-llms.md>), [reasoning](<https://devfeed.tech/tags/reasoning.md>), [scaling](<https://devfeed.tech/tags/scaling.md>), [supervision](<https://devfeed.tech/tags/supervision.md>)

### AI overview

CauSim makes causal reasoning supervision more scalable by using LLMs to build increasingly complex executable structural causal models (SCMs) with verifiable answers to causal queries. It formalizes non-executable causal knowledge as code to enable data augmentation, then translates SCMs into natural language for supervision across representations. The abstract reports cross-representation generalization, gains from curriculum scaling and more data, LLM self-improvement using self-generated simulators, and augmentation from existing domain knowledge.

### Source excerpt

arXiv:2605.09079v2 Announce Type: replace Abstract: Despite surpassing human performance across mathematics, coding, and other knowledge-intensive tasks, large language models (LLMs) continue to struggle with causal reasoning. A core obstacle is the target data itself: causal systems are complex and often expressed in non-executable forms, while ground-truth answers to causal queries are inherently scarce. We introduce CauSim, a framework that turns causal reasoning from a scarce-label problem into a scalable supervised one. CauSim constructs increasingly complex causal simulators: executable structural causal models (SCMs), incrementally built by LLMs, that scale to globally complex systems while maintaining verifiable answers to causal queries. CauSim operates across representations by formalizing non-executable causal knowledge into code, enabling data augmentation, and translating executable SCMs into natural language, enabling supervision in previously difficult-to-supervise representations. We structure our research into two parts: (1) how to construct increasingly complex causal simulators, and (2) a systematic study of what CauSim enables, demonstrating generalization across representations, consistent gains from curriculum scaling and data volume, LLM self-improvement through self-generated simulators, and data augmentation via formalization of existing domain knowledge.

## From Constraints to Causality: Petri net Encodings for Explaining Binary Neural Network Inference

DevFeed: [From Constraints to Causality: Petri net Encodings for Explaining Binary Neural Network Inference](<https://devfeed.tech/articles/from-constraints-to-causality-petri-net-encodings-for-explaining-binary-neural-network-inference-141771.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.22396>)

Author: Alex Chan (Newcastle University), Mohamed Tarraf (Newcastle University), Alex Yakovlev (Newcastle University), Rishad Shafik (Newcastle University)

Published: 2026-11-18T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Neural Network](<https://devfeed.tech/topics/neural-network.md>), [causality](<https://devfeed.tech/topics/causality.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>)

Tags: [causality](<https://devfeed.tech/tags/causality.md>), [constraint](<https://devfeed.tech/tags/constraint.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-lo](<https://devfeed.tech/tags/cs-lo.md>), [decision](<https://devfeed.tech/tags/decision.md>), [inference](<https://devfeed.tech/tags/inference.md>), [neural](<https://devfeed.tech/tags/neural.md>), [neural-network](<https://devfeed.tech/tags/neural-network.md>), [safety](<https://devfeed.tech/tags/safety.md>), [semantics](<https://devfeed.tech/tags/semantics.md>), [sequence](<https://devfeed.tech/tags/sequence.md>), [trace](<https://devfeed.tech/tags/trace.md>), [translation](<https://devfeed.tech/tags/translation.md>)

### AI overview

The paper presents a compositional translation of a fixed, trained Binary Neural Network's inference into a 1-safe Petri net, making each computation step an explicit transition and each firing sequence a causal trace of a decision. This offers a complementary view to SAT and SMT verification, which the abstract says establishes input-output properties but does not show how the computation reaches its result. In a worked example, exhaustive input-output comparison with a reference implementation supports faithfulness, and contrasting traces show where changing one feature alters the prediction. The authors propose future work on proving translation correctness and tracing where two trained BNNs diverge; these are research directions, not results established in the excerpt.

### Source excerpt

arXiv:2611.22396v1 Announce Type: cross Abstract: Binary Neural Networks (BNNs) are attractive for resource-constrained hardware and on-device components in autonomous systems. However, their discrete, non-linear behaviour makes them difficult to analyse and trust in safety-relevant settings. Existing approaches based on SAT and Satisfiability Modulo Theories can verify a BNN's input-output properties, but treat it as a monolithic constraint system, which reveal only whether a property holds and not how the computation reaches its result. In this paper, we present an early work that provides a complementary, discrete-event view. We give a compositional translation from the inference semantics of a fixed, trained BNN to a 1-safe Petri net (PN), where each computation step is an explicit transition, so that a firing sequence is a causal trace of the BNN's decision. On a worked example, we establish faithfulness via exhaustive input-output equivalence against a reference implementation, and show how contrasting firing sequences localises where a single-feature change alters the prediction. Finally, we outline two directions this opens: a correctness-of-translation theorem, and a causal-trace analysis of where two trained BNNs diverge in their decisions, a structural question that constraint-based verifiers cannot express.

## Symmetries of Counterfactuals: Identification and Testability of Counterfactual Maps from Covariate Heterogeneity

DevFeed: [Symmetries of Counterfactuals: Identification and Testability of Counterfactual Maps from Covariate Heterogeneity](<https://devfeed.tech/articles/symmetries-of-counterfactuals-identification-and-testability-of-counterfactual-maps-from-covariate-heterogeneity-141823.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.22948>)

Author: Marc Braun, Adel Daoud, Jose M. Pe\\~na

Published: 2026-11-18T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [causality](<https://devfeed.tech/topics/causality.md>)

Tags: [all](<https://devfeed.tech/tags/all.md>), [conditional](<https://devfeed.tech/tags/conditional.md>), [consistency](<https://devfeed.tech/tags/consistency.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [data](<https://devfeed.tech/tags/data.md>), [distributions](<https://devfeed.tech/tags/distributions.md>), [maps](<https://devfeed.tech/tags/maps.md>), [mechanism](<https://devfeed.tech/tags/mechanism.md>), [models](<https://devfeed.tech/tags/models.md>), [optimal-transport](<https://devfeed.tech/tags/optimal-transport.md>), [stat-ml](<https://devfeed.tech/tags/stat-ml.md>), [statistics](<https://devfeed.tech/tags/statistics.md>), [synthetic](<https://devfeed.tech/tags/synthetic.md>), [test](<https://devfeed.tech/tags/test.md>)

### AI overview

Counterfactual maps can remain unidentified even when observed distributions are known. This paper characterizes the ambiguity as symmetries that preserve conditional outcome distributions given covariates that affect the outcome only through its latent cause; greater variation in those distributions across covariate values yields a smaller symmetry group. For triangular monotone and optimal transport map classes, the authors show that consistency with the data is equivalent to a single conditional independence, providing a test of the assumed class. The abstract says they illustrate the results with synthetic data and the ACTG 175 HIV trial.

### Source excerpt

arXiv:2611.22948v1 Announce Type: cross Abstract: In general, counterfactuals are not identified from data alone, because many causal models produce the same observed distributions but different counterfactuals. A common approach to circumvent this problem is to restrict the mechanism to a class in which it becomes unique, such as triangular monotone maps or optimal transport maps. However, whether the assumed class is correct is usually not examined. In this work, we study what the data can tell us when covariates are available that influence the outcome only through its latent cause. First, we characterise all counterfactual maps that are consistent with the data. They differ from the true map by a group of symmetries that preserve the conditional outcome distributions given the covariates. The more these distributions vary across covariate values, the smaller the group. Second, we show that for the classes above, consistency with the data is equivalent to a single conditional independence, which yields a test of the assumed class. We illustrate both results on synthetic data and on the ACTG 175 HIV trial.

## What MoE Routing Controls: Causal Interventions on Expert Trajectories

DevFeed: [What MoE Routing Controls: Causal Interventions on Expert Trajectories](<https://devfeed.tech/articles/what-moe-routing-controls-causal-interventions-on-expert-trajectories-141735.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.22120>)

Author: Mukund Athmaram, Manoj Rao, Liang Ma

Published: 2026-11-18T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Mixture of Experts (MoE)](<https://devfeed.tech/topics/mixture-of-experts-moe.md>), [causality](<https://devfeed.tech/topics/causality.md>), [Language models](<https://devfeed.tech/topics/language-models.md>)

Tags: [analysis](<https://devfeed.tech/tags/analysis.md>), [bias](<https://devfeed.tech/tags/bias.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [experts](<https://devfeed.tech/tags/experts.md>), [form](<https://devfeed.tech/tags/form.md>), [format](<https://devfeed.tech/tags/format.md>), [formats](<https://devfeed.tech/tags/formats.md>), [generation](<https://devfeed.tech/tags/generation.md>), [language](<https://devfeed.tech/tags/language.md>), [language-models](<https://devfeed.tech/tags/language-models.md>), [layers](<https://devfeed.tech/tags/layers.md>), [mixture-of-experts](<https://devfeed.tech/tags/mixture-of-experts.md>), [mixture-of-experts-moe](<https://devfeed.tech/tags/mixture-of-experts-moe.md>), [model](<https://devfeed.tech/tags/model.md>), [models](<https://devfeed.tech/tags/models.md>), [moe](<https://devfeed.tech/tags/moe.md>), [money](<https://devfeed.tech/tags/money.md>), [qwen3](<https://devfeed.tech/tags/qwen3.md>), [religion](<https://devfeed.tech/tags/religion.md>), [router](<https://devfeed.tech/tags/router.md>), [routing](<https://devfeed.tech/tags/routing.md>)

### AI overview

The study tests how steering mixture-of-experts routing affects generation in Qwen3-30B-A3B. Researchers build expert paths from tokens associated with a format or concept, then bias the corresponding router logits during generation. Steering format paths changes the produced text; steering concept paths shifts next-token probabilities toward the concept, including for paths built in another language, but has little effect on generated text. Partial-generation interventions suggest that a format can reinforce itself once begun. Religion and money show the strongest concept steering, with effects concentrated in middle layers. The excerpt says the authors examine embedding distinctiveness and word frequency to explain variation across concepts, but does not report the results of that analysis.

### Source excerpt

arXiv:2611.22120v1 Announce Type: cross Abstract: In mixture-of-experts (MoE) language models, a token is routed to a sequence of experts across layers. We investigate how forcing tokens onto chosen expert paths influences model behavior. We build a target path from the experts that a concept's or role's tokens most often use, and we add a fixed bias to each target expert's router logit during generation to steer the token toward this expert path. We study two kinds of targets. Format targets concern the form of the output, and concept targets concern its topic. In Qwen3-30B-A3B, we show that steering a format path changes the text the model writes. We also show that steering a concept path shifts the model's next-token probabilities toward the concept, even when the path is built in another language, yet this shift has little effect on the generated text. We investigate why formats reach the text but concepts do not by steering only part of the generation, and find evidence that a format, once started, reinforces itself. We show that the concepts religion and money steer most strongly, and that the effect is concentrated in the middle layers. We attempt to explain why the strength of the effect varies by concept through an analysis of embedding distinctiveness and word frequency.

## RAW: Refusal Aversion in the Wild, A Causal Measurement Method for Deployed LLMs

DevFeed: [RAW: Refusal Aversion in the Wild, A Causal Measurement Method for Deployed LLMs](<https://devfeed.tech/articles/raw-refusal-aversion-in-the-wild-a-causal-measurement-method-for-deployed-llms-139886.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.21641>)

Author: Pavel Kireyev

Published: 2026-11-18T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Large Language Model](<https://devfeed.tech/topics/llm.md>), [causality](<https://devfeed.tech/topics/causality.md>)

Tags: [a-b-testing](<https://devfeed.tech/tags/a-b-testing.md>), [code](<https://devfeed.tech/tags/code.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [cs-hc](<https://devfeed.tech/tags/cs-hc.md>), [data](<https://devfeed.tech/tags/data.md>), [econ-gn](<https://devfeed.tech/tags/econ-gn.md>), [engagement](<https://devfeed.tech/tags/engagement.md>), [experiment](<https://devfeed.tech/tags/experiment.md>), [llms](<https://devfeed.tech/tags/llms.md>), [logs](<https://devfeed.tech/tags/logs.md>), [measurement](<https://devfeed.tech/tags/measurement.md>), [q-fin-ec](<https://devfeed.tech/tags/q-fin-ec.md>)

### AI overview

RAW estimates refusal's effect on subsequent user re-engagement from existing conversation logs, using variation in whether the same deployed model refuses near-identical prompts as a natural experiment. Applied to 15 models in LMSYS-Chat-1M and WildChat with user-clustered inference, it finds a measurable 3-17 percentage-point reduction for about half of the models measured precisely; the others show no distinguishable effect. One deployment's positive point estimate is attributed to persistent users retrying refused requests. The method requires no refusal A/B test and is reported to cost about $0.30 per 1,000 conversations. The supplied abstract describes the method and results but does not provide further methodological details.

### Source excerpt

arXiv:2611.21641v1 Announce Type: cross Abstract: Deployed LLMs field hundreds of millions of conversations a day and refuse a non-trivial fraction of user requests, yet the user-side cost of refusal, whether users disengage and by how much, is unmeasured on any production model: a blind spot for the consumer-facing platforms built on them. A/B-testing refusal is ethically and operationally constrained on frontier-lab products and unavailable to third-party platforms deploying these models through a fixed API. We introduce RAW, a method that uses LLM stochasticity as a natural experiment: at near-identical prompts in any existing log, the same model sometimes refuses and sometimes does not, and that variation identifies the causal user-side effect of refusal. Applying the method to 15 deployed LLMs in LMSYS-Chat-1M and WildChat, with user-clustered inference, we find that for about half of the models we can measure precisely, a refusal costs 3 to 17 pp of subsequent re-engagement; the rest show no distinguishable effect, among them one deployment whose positive point estimate we trace to persistent users retrying the refused request. Our contribution is the measurement method, which runs on any deployed model from logs alone for about \$0.30 per 1,000 conversations. We release labels, moderation scores, prompt embeddings, and pipeline code.

## UrbanROI: Causal Modeling for Regional ROI Forecasting under POI Entry

DevFeed: [UrbanROI: Causal Modeling for Regional ROI Forecasting under POI Entry](<https://devfeed.tech/articles/urbanroi-causal-modeling-for-regional-roi-forecasting-under-poi-entry-139004.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.20180>)

Author: Zhaoqi Zhang, Miao Xie, Linyou Cai, Qianlong Xie, Xingxing Wang, Siqiang Luo, Gao Cong

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [causality](<https://devfeed.tech/topics/causality.md>), [Data analysis](<https://devfeed.tech/topics/data-analysis.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [ads](<https://devfeed.tech/tags/ads.md>), [advertising](<https://devfeed.tech/tags/advertising.md>), [analysis](<https://devfeed.tech/tags/analysis.md>), [business](<https://devfeed.tech/tags/business.md>), [competition](<https://devfeed.tech/tags/competition.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [datasets](<https://devfeed.tech/tags/datasets.md>), [decision](<https://devfeed.tech/tags/decision.md>), [decision-making](<https://devfeed.tech/tags/decision-making.md>), [effects](<https://devfeed.tech/tags/effects.md>), [evolution](<https://devfeed.tech/tags/evolution.md>), [forecasting](<https://devfeed.tech/tags/forecasting.md>), [framework](<https://devfeed.tech/tags/framework.md>), [future](<https://devfeed.tech/tags/future.md>), [graph](<https://devfeed.tech/tags/graph.md>), [graphs](<https://devfeed.tech/tags/graphs.md>), [impact](<https://devfeed.tech/tags/impact.md>), [interaction](<https://devfeed.tech/tags/interaction.md>), [modeling](<https://devfeed.tech/tags/modeling.md>), [roi](<https://devfeed.tech/tags/roi.md>)

### AI overview

UrbanROI forecasts how a commercial region's ROI trajectory may change after a new point of interest (POI) enters. It addresses a gap in methods focused on individual ads or merchants by modeling temporal ROI and competition among surrounding POIs. Its Interaction-aware Regional Graph Module uses user request contexts, spatial proximity, and request-level commercial signals to represent competitive relationships; structurally aligned treatment and control graphs simulate factual and counterfactual regional states. The abstract reports that experiments on large-scale real-world POI and advertising datasets found UrbanROI consistently outperformed strong baselines in post-entry regional ROI forecasting.

### Source excerpt

arXiv:2611.20180v1 Announce Type: new Abstract: As urban commercial environments continue to evolve rapidly, accurately modeling the economic impact of point of interest (POI) entry is essential for supporting data-driven advertising allocation, merchant onboarding, and regional business decision-making. Existing return on investment (ROI) prediction methods mainly focus on individual ads or merchants, overlooking the interaction effects among POIs and failing to capture the causal influence of structural changes in urban commercial ecosystems. In this paper, we introduce a novel research problem -- regional ROI forecasting under POI entry, which aims to predict the future ROI trajectory of a commercial region after the introduction of a new POI by modeling temporal ROI evolution and competition-driven interactions among surrounding POIs. To address these challenges, we propose UrbanROI, an intervention-aware causal forecasting framework for region-level economic outcome prediction. UrbanROI leverages an Interaction-aware Regional Graph Module to explicitly model competitive relationships among POIs based on user request contexts, spatial proximity, and request-level commercial signals, and constructs structurally aligned treatment and control graphs to simulate factual and counterfactual regional states. Extensive experiments on large-scale real-world POI and advertising datasets demonstrate that UrbanROI consistently outperforms strong baselines in post-entry regional ROI forecasting. The results demonstrate its effectiveness in temporal economic modeling, inter-POI interaction representation, and intervention-aware forecasting, providing support for regional economic analysis and business decision-making in dynamic urban commercial systems.

## Contextual Causal Bayesian Optimisation

DevFeed: [Contextual Causal Bayesian Optimisation](<https://devfeed.tech/articles/contextual-causal-bayesian-optimisation-138703.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2301.12412>)

Author: Vahan Arsenyan, Antoine Grosnit, Haitham Bou-Ammar, Arnak Dalalyan

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [causality](<https://devfeed.tech/topics/causality.md>), [Hyperparameter optimization](<https://devfeed.tech/topics/hyperparameter-optimization.md>), [Context-aware AI](<https://devfeed.tech/topics/context-aware-ai.md>)

Tags: [algorithm](<https://devfeed.tech/tags/algorithm.md>), [approach](<https://devfeed.tech/tags/approach.md>), [bayesian](<https://devfeed.tech/tags/bayesian.md>), [complexity](<https://devfeed.tech/tags/complexity.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [design](<https://devfeed.tech/tags/design.md>), [environments](<https://devfeed.tech/tags/environments.md>), [expectation](<https://devfeed.tech/tags/expectation.md>), [experimental](<https://devfeed.tech/tags/experimental.md>), [framework](<https://devfeed.tech/tags/framework.md>), [graph](<https://devfeed.tech/tags/graph.md>), [information](<https://devfeed.tech/tags/information.md>), [novel](<https://devfeed.tech/tags/novel.md>), [optimisation](<https://devfeed.tech/tags/optimisation.md>), [policies](<https://devfeed.tech/tags/policies.md>), [probability](<https://devfeed.tech/tags/probability.md>), [results](<https://devfeed.tech/tags/results.md>), [sample](<https://devfeed.tech/tags/sample.md>), [search](<https://devfeed.tech/tags/search.md>), [target](<https://devfeed.tech/tags/target.md>), [variable](<https://devfeed.tech/tags/variable.md>)

### AI overview

The proposed framework combines contextual information with known causal graphs to optimize intervention policies for a target variable. Its algorithm jointly selects the policy and the variables on which that policy depends, unifying contextual and causal Bayesian optimization to address cases where either approach alone can yield suboptimal results. The authors derive worst-case and instance-dependent high-probability regret bounds, and report experiments showing sublinear regret and reduced sample complexity in high-dimensional settings.

### Source excerpt

arXiv:2301.12412v5 Announce Type: replace Abstract: We introduce a unified framework for contextual and causal Bayesian optimisation, which aims to design intervention policies maximising the expectation of a target variable. Our approach leverages both observed contextual information and known causal graph structures to guide the search. Within this framework, we propose a novel algorithm that jointly optimises over policies and the sets of variables on which these policies are defined. This thereby extends and unifies two previously distinct approaches: Causal Bayesian Optimisation and Contextual Bayesian Optimisation, while also addressing their limitations in scenarios that yield suboptimal results. We derive worst-case and instance-dependent high-probability regret bounds for our algorithm. We report experimental results across diverse environments, corroborating that our approach achieves sublinear regret and reduces sample complexity in high-dimensional settings.

## Pretty Linear Liars: Unregularized Probes Favor Intervention-Unfaithful Features

DevFeed: [Pretty Linear Liars: Unregularized Probes Favor Intervention-Unfaithful Features](<https://devfeed.tech/articles/pretty-linear-liars-unregularized-probes-favor-intervention-unfaithful-features-137565.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.20371>)

Author: Karl Vilhelmsson Emneby, Julian Yocum, Cameron Allen

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [linear-regression](<https://devfeed.tech/topics/linear-regression.md>), [causality](<https://devfeed.tech/topics/causality.md>)

Tags: [causality](<https://devfeed.tech/tags/causality.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [discovery](<https://devfeed.tech/tags/discovery.md>), [features](<https://devfeed.tech/tags/features.md>), [interpretability](<https://devfeed.tech/tags/interpretability.md>), [linear-probing](<https://devfeed.tech/tags/linear-probing.md>), [llama](<https://devfeed.tech/tags/llama.md>), [regression](<https://devfeed.tech/tags/regression.md>), [transformer](<https://devfeed.tech/tags/transformer.md>), [variance](<https://devfeed.tech/tags/variance.md>)

### AI overview

The study finds that unregularized OLS probes can favor features that are decodable but do not faithfully reflect the model's active use of them: theoretically, OLS amplifies low-variance activation directions spuriously correlated with the target. In experiments with a toy transformer and Llama-2-7B, intervention faithfulness rises with ridge regularization to an optimum, then falls as regularization increases further. At that optimum, probe features resemble those learned by directly optimizing intervention outcomes in both direction and variance. The results support regularizing regression probes, while indicating that the strength matters.

### Source excerpt

arXiv:2611.20371v1 Announce Type: cross Abstract: Linear probing is a common technique in mechanistic interpretability despite the fact that the decodability of a given feature does not necessarily imply the network uses that feature. Intervention faithfulness, not merely decodability, is the appropriate criterion for whether a feature serves a defined active role in the model. Meanwhile, probe regularization has been used for feature discovery, but the impact of regularization on feature causality remains unclear. We study probe regularization systematically, starting with unregularized ordinary least squares (OLS) regression and considering increasing amounts of regularization via ridge regression. We show theoretically that OLS probes amplify low-variance directions in activation space that have spurious correlations with the target feature, whereas regularization reduces influence from these directions. Empirically, we find in a toy transformer and Llama-2-7B that intervention faithfulness increases up to a point, and then decreases again, and that the optimal regularization strength leads to features bearing striking similarity (in both direction and variance) to those learned by optimizing for intervention outcomes directly. Together, these results provide mechanistic justification for regularizing regression probes.

## Multi-Objective Causal Bandits: Minimal Intervention Space and Policy-Level Learning

DevFeed: [Multi-Objective Causal Bandits: Minimal Intervention Space and Policy-Level Learning](<https://devfeed.tech/articles/multi-objective-causal-bandits-minimal-intervention-space-and-policy-level-learning-138900.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.19634>)

Author: Muhammad Qasim Elahi, Mahsa Ghasemi, Murat Kocaoglu

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [causality](<https://devfeed.tech/topics/causality.md>), [Reinforcement learning](<https://devfeed.tech/topics/reinforcement-learning.md>)

Tags: [algorithm](<https://devfeed.tech/tags/algorithm.md>), [bandits](<https://devfeed.tech/tags/bandits.md>), [comparisons](<https://devfeed.tech/tags/comparisons.md>), [constraint](<https://devfeed.tech/tags/constraint.md>), [convex](<https://devfeed.tech/tags/convex.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [graph](<https://devfeed.tech/tags/graph.md>), [guarantees](<https://devfeed.tech/tags/guarantees.md>), [has](<https://devfeed.tech/tags/has.md>), [learning](<https://devfeed.tech/tags/learning.md>), [minimal](<https://devfeed.tech/tags/minimal.md>), [objectives](<https://devfeed.tech/tags/objectives.md>), [online-learning](<https://devfeed.tech/tags/online-learning.md>), [optimize](<https://devfeed.tech/tags/optimize.md>), [pareto-optimality](<https://devfeed.tech/tags/pareto-optimality.md>), [performance](<https://devfeed.tech/tags/performance.md>), [policies](<https://devfeed.tech/tags/policies.md>), [policy](<https://devfeed.tech/tags/policy.md>), [reward](<https://devfeed.tech/tags/reward.md>), [safety](<https://devfeed.tech/tags/safety.md>), [space](<https://devfeed.tech/tags/space.md>)

### AI overview

The paper shows that causal bandit actions can be suboptimal because a randomized policy over other interventions may dominate them, even when no single intervention does. It characterizes, from the causal graph, a minimal and complete family of intervention sets that could be Pareto-optimal, and reports this corrects earlier formulations that could include sets never optimal under any compatible structural causal model. On this reduced space, its UCB-style algorithm tests dominance against convex combinations and achieves logarithmic gap-dependent Pareto regret. For constrained settings, where feasibility or optimality may require randomization, it presents a UCB-style algorithm with sublinear regret and constraint-violation guarantees.

### Source excerpt

arXiv:2611.19634v1 Announce Type: new Abstract: Many decision-making problems involve multiple objectives, where actions affect several performance metrics, fairness criteria, or safety constraints. We study multi-objective causal bandits with multiple reward nodes, where actions correspond to interventions on subsets of variables in a causal graph and performance is evaluated through Pareto optimality. Existing approaches to multi-objective bandits often rely on pairwise Pareto dominance or optimize a fixed scalarization. However, pairwise comparisons alone can fail to identify actions dominated by randomized policies, whose expected reward vectors lie in the convex hull of action-level rewards. Thus, an intervention may be suboptimal even if no single intervention dominates it, because it can be dominated by a randomized policy over other interventions. We formalize this policy-level phenomenon and show that it has consequences for both causal action-space reduction and online learning. First, we provide a necessary-and-sufficient graphical characterization of possibly Pareto- optimal minimal intervention sets. Our characterization yields a minimal, sound, and complete candidate intervention family from the graph alone, correcting prior formulations that may include intervention sets that are never Pareto-optimal under any compatible structural causal model. Second, over this reduced intervention space, we design an online UCB-style algorithm that eliminates actions using dominance tests against convex combinations of other actions, and prove logarithmic gap-dependent Pareto regret. Finally, we study constrained multi-objective causal bandits, where feasibility and optimality may be realized only with randomized policies, and develop an online UCB-style constrained bandit algorithm with sublinear regret and constraint-violation guarantees.

## Certified Policy Optimisation for Nested Causal Bandits via PAC-Bayes Risk

DevFeed: [Certified Policy Optimisation for Nested Causal Bandits via PAC-Bayes Risk](<https://devfeed.tech/articles/certified-policy-optimisation-for-nested-causal-bandits-via-pac-bayes-risk-134987.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2605.29788>)

Author: Tim Woydt, Paul-David Zuercher

Published: 2026-11-16T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [thompson sampling](<https://devfeed.tech/topics/thompson-sampling.md>), [Reinforcement learning](<https://devfeed.tech/topics/reinforcement-learning.md>), [causality](<https://devfeed.tech/topics/causality.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [bandit](<https://devfeed.tech/tags/bandit.md>), [bandits](<https://devfeed.tech/tags/bandits.md>), [bayes](<https://devfeed.tech/tags/bayes.md>), [bayesian](<https://devfeed.tech/tags/bayesian.md>), [certificate](<https://devfeed.tech/tags/certificate.md>), [context](<https://devfeed.tech/tags/context.md>), [controller](<https://devfeed.tech/tags/controller.md>), [coupling](<https://devfeed.tech/tags/coupling.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [data](<https://devfeed.tech/tags/data.md>), [decision](<https://devfeed.tech/tags/decision.md>), [deployment](<https://devfeed.tech/tags/deployment.md>), [distribution](<https://devfeed.tech/tags/distribution.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [nested](<https://devfeed.tech/tags/nested.md>), [optimisation](<https://devfeed.tech/tags/optimisation.md>), [policy](<https://devfeed.tech/tags/policy.md>), [regression](<https://devfeed.tech/tags/regression.md>), [risk](<https://devfeed.tech/tags/risk.md>)

### AI overview

The paper models nested strategic and tactical decisions as Nested Contextual Causal Bandits (NCCBs), where an action at one level changes the next level's context distribution. Its Nested Causal Thompson Sampling (NCTS) draws a mechanism-factorised belief once per episode and acts recursively. A causal PAC-Bayesian excess-risk bound can certify a candidate policy using historical data, off-policy and at any time. In hierarchical SCM experiments, the factorised posterior transferred better zero-shot under exogenous shifts than matched RFF-GP joint regression; recursive meta-to-inner commits outperformed joint commits in distribution, and certificates tightened as offline data grew. The authors use this to propose progressive certified handover, switching timescales independently from legacy controllers to NCTS when gains are certifiable.

### Source excerpt

arXiv:2605.29788v2 Announce Type: replace Abstract: Critical sequential decisions are rarely single-timescale: a strategic decision causally shapes the context in which every subsequent tactical choice is made; standard bandit and reinforcement-learning theory does not capture this causal coupling between timescales. We formalise the problem class as Nested Contextual Causal Bandits (NCCBs), a hierarchical SCM where each level's action sets the next level's context distribution, and propose Nested Causal Thompson Sampling (NCTS), which draws one mechanism-factorised belief per episode and acts recursively under it. Our main theoretical result is a causal PAC-Bayesian excess-risk bound that certifies any candidate deployment policy from historic data alone, off-policy and anytime, answering the deployment question: can we trust this agent here, and at what risk? Experiments on a hierarchical SCM show that, against a matched RFF-GP joint regression on the same function class, the factorised SCM-mechanism posterior transfers significantly better zero-shot under exogenous distribution shifts, the recursive meta-to-inner commit significantly dominates the joint-commit alternative in distribution, and the certificate significantly contracts as offline data accumulates. Combining these results, we establish progressive certified handover, a safe-deployment method: each timescale flips from a legacy controller to NCTS when gains can be certified, independently of the others.

## Efficient and Practical Clustering of Variables in Causal Graphs

DevFeed: [Efficient and Practical Clustering of Variables in Causal Graphs](<https://devfeed.tech/articles/efficient-and-practical-clustering-of-variables-in-causal-graphs-135851.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.17971>)

Author: Otto Tabell, Jani Nyk\\"anen, Santtu Tikka, Juha Karvanen

Published: 2026-11-16T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [clustering](<https://devfeed.tech/topics/clustering.md>), [causality](<https://devfeed.tech/topics/causality.md>), [graph theory](<https://devfeed.tech/topics/graph-theory.md>)

Tags: [algorithm](<https://devfeed.tech/tags/algorithm.md>), [applications](<https://devfeed.tech/tags/applications.md>), [causal-inference](<https://devfeed.tech/tags/causal-inference.md>), [clustering](<https://devfeed.tech/tags/clustering.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [graphs](<https://devfeed.tech/tags/graphs.md>), [real-world](<https://devfeed.tech/tags/real-world.md>), [reduction](<https://devfeed.tech/tags/reduction.md>), [simulation](<https://devfeed.tech/tags/simulation.md>), [stat-me](<https://devfeed.tech/tags/stat-me.md>), [stat-ml](<https://devfeed.tech/tags/stat-ml.md>), [variables](<https://devfeed.tech/tags/variables.md>)

### AI overview

The paper introduces "conduits," a class of clusters for variables in causal graphs, and a sound, complete, efficient algorithm to detect all such clusters. The authors report that conduits occur naturally in real-world causal graphs and can substantially reduce graph size. Their structure also permits applying existing identifiability results for clustering. A simulation study using real-world graphs demonstrates these properties.

### Source excerpt

arXiv:2611.17971v1 Announce Type: cross Abstract: Clustering variables in a graph may improve visual clarity and simplify causal inference. Harnessing the full potential of clustering requires an automated method to detect useful clusters in a graph. We introduce a novel class of clusters called conduits, and present a sound, complete and efficient algorithm for detecting all conduits in the graph. In addition to algorithmic detectability, conduits appear naturally in causal graphs stemming from real-world applications, enabling substantial size reduction. Furthermore, the structure of conduits enables the application of known identifiability results regarding clustering. We demonstrate these properties in a simulation study with real-world graphs.

## Tracing Actual Causes with Counterfactual Witness Maps

DevFeed: [Tracing Actual Causes with Counterfactual Witness Maps](<https://devfeed.tech/articles/tracing-actual-causes-with-counterfactual-witness-maps-135150.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.17937>)

Author: Tim Woydt, Matej Ze\\v{c}evi\\'c, Kristian Kersting, Moritz Willig

Published: 2026-11-16T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [causality](<https://devfeed.tech/topics/causality.md>)

Tags: [algorithm](<https://devfeed.tech/tags/algorithm.md>), [all](<https://devfeed.tech/tags/all.md>), [assessment](<https://devfeed.tech/tags/assessment.md>), [attribution](<https://devfeed.tech/tags/attribution.md>), [boolean](<https://devfeed.tech/tags/boolean.md>), [cause](<https://devfeed.tech/tags/cause.md>), [compare](<https://devfeed.tech/tags/compare.md>), [complete](<https://devfeed.tech/tags/complete.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-lo](<https://devfeed.tech/tags/cs-lo.md>), [enumeration](<https://devfeed.tech/tags/enumeration.md>), [event](<https://devfeed.tech/tags/event.md>), [events](<https://devfeed.tech/tags/events.md>), [existing](<https://devfeed.tech/tags/existing.md>), [formal](<https://devfeed.tech/tags/formal.md>), [framework](<https://devfeed.tech/tags/framework.md>), [full](<https://devfeed.tech/tags/full.md>), [gap](<https://devfeed.tech/tags/gap.md>), [grammar](<https://devfeed.tech/tags/grammar.md>), [graphs](<https://devfeed.tech/tags/graphs.md>), [maps](<https://devfeed.tech/tags/maps.md>), [tracing](<https://devfeed.tech/tags/tracing.md>)

### AI overview

TrAC addresses the lack of a sound and complete procedure for enumerating all actual causes of an event in an acyclic structural causal model. It combines witness targets, which contain every counterfactual witness, with an ancestor intervention grammar to construct a finite directed acyclic witness map. The authors prove that each actual cause is represented by a node for some contingency set, and report that experiments on random Boolean models show practical feasibility up to 15 nodes. The abstract does not provide comparative performance details beyond comparison with existing single-cause baselines.

### Source excerpt

arXiv:2611.17937v1 Announce Type: new Abstract: Halpernian actual causation, the formal study of which events in a specific situation caused a specific outcome in acyclic structural causal models, underpins legal responsibility, moral blame, and the assessment of causal harm. Despite the rich logical framework, no prior sound and complete enumeration procedure is known for finding all actual causes in a given setting, blocking downstream tasks such as responsibility attribution, blame assessment, and causal-harm grading that aggregate over the full cause set. We close this gap by introducing witness targets: sets of joint configurations containing every counterfactual witness. In combination with an ancestor intervention grammar we derive a finite directed acyclic witness map. We prove that for every actual cause there exists a contingency set such that the joint intervention set appears as a node of this map, and present an algorithm for Tracing Actual Causes (TrAC) that is sound and complete for enumerating all actual causes of any target event. We compare TrAC empirically with existing single-cause baselines on random Boolean structural causal models, demonstrating practical feasibility for graphs of up to 15 nodes.

[Next page](<https://devfeed.tech/topics/causality.md?cursor=WyIyMDI2LTExLTE2VDA1OjAwOjAwKzAwOjAwIiwgIjI1N2M1MTFhLThmOWYtNDI5OS1iMDI3LTM1OGIwYzVkOWZmMSJd>)