arXiv Machine Learning
Jul 7
Identifiability Without Gaussianity: Symbolic World Models and Near-Infinite Temporal Consistency
arXiv:2606. 12471v2 Announce Type: replace-cross Abstract: Klindt, LeCun, and Balestriero (arXiv:2605.
WorldKernel: A World Model is the Coupling Kernel of Admissible Possible Worlds
arXiv:2606. 10934v1 Announce Type: new Abstract: A common assumption holds that enough observational and interventional data, given to a strong enough predictor, suffices.
Related stories
arXiv Machine Learning
Aug 26
Bayes with No Shame: Admissibility Geometries of Predictive Inference
arXiv:2603.05335v3 Announce Type: replace-cross Abstract: Modern predictive systems combine predictors, sequential monitors, prediction sets, and online strategies, each with a different certificate...
arXiv AI
Jul 20
A cubical formalisation of topos causal models: intervention, sheaf gluing, and the intuitionistic do-calculus
arXiv:2607. 15629v1 Announce Type: cross Abstract: Topos causal models recast causal inference inside a topos: a causal world is a presheaf, an intervention is a characteristic map into the subobject classifier, and reasoning is carried out in the intuitionistic internal language.
arXiv AI
Aug 24
World models of environment, agent and joint agent-environment systems
The paper introduces a framework that distinguishes world models by the channel they represent—environment, agent, or joint agent‑environment—using computational mechanics to define canonical predictive models as ε-transducers or ε-machines. It shows how closed‑loop coupling induces support‑restricted models whose states factor through the joint causal states, and demonstrates with a POMDP example that such restriction can reduce an otherwise infinite‑state environment model to a finite one.
arXiv Machine Learning
Sep 4
Guide, Not Bind: Why Defeasible Priors Fail in Augmented Lagrangian Causal Discovery
The paper investigates why differentiable causal discovery methods that encode expert priors as forbidden-edge constraints via an Augmented Lagrangian (ALM) penalty—termed the "guide, not bind" approach—often fail. It identifies two key failures: (1) the sequential penalty‑ramping ALM suppresses a true edge before counterfactual checks can detect it, and the proposed adaptive relaxation rule DADU violates necessary conditions for safe relaxation, leading to a high failure rate across thousands of training runs; (2) the standard correlation‑matching objective inherently ties a true edge and its reverse to the same cost, whereas covariance matching can separate them by a provable margin. The authors provide theoretical propositions, corollaries, and empirical evidence to support these claims.
arXiv Machine Learning
Aug 18
Generalised Transportability via Causal Abstractions
arXiv:2608. 15645v1 Announce Type: new Abstract: Transporting a causal conclusion from a source study population to a target one is a fundamental problem in causal inference.
arXiv AI
Aug 11
CausalNav: Reliability-Certified Causal World Models for Control under Physical-Parameter Shift
arXiv:2608. 07809v1 Announce Type: new Abstract: A world model is only useful for physical AI if it changes what the agent does, and only safe if it declines to do so when it is wrong.
arXiv AI
Jun 4
Trivium: Temporal Regret as a First-Class Objective for Causal-Memory Controllers
arXiv:2606. 04421v1 Announce Type: new Abstract: Many current agentic systems and LLM pipelines correct mistakes by optimizing outcome reward.
arXiv Machine Learning
Aug 17
Identifiability and Order-Dimension Limits of In-Context Learning on Partial Orders
arXiv:2608. 14004v1 Announce Type: new Abstract: In-context learning is commonly formalized as inference from examples of a function.
arXiv Machine Learning
Sep 23
PICPIs: Prediction-Interval-Conditional Prediction Intervals
arXiv:2609.25388v1 Announce Type: cross Abstract: A classical question in statistics is which observable quantities to condition on when drawing inferences about unobservable targets. For conformal p...
arXiv AI
Aug 18
Constitutive Priors for Machine Intelligence: A Legitimacy Theory of the Artificial Physical World
arXiv:2608. 15147v1 Announce Type: new Abstract: Machine intelligence has conquered the symbolic world but stalled at the physical one.
arXiv AI
Sep 4
The Geometry of Ignorance: LLMs Know When to Temper Bayesian Priors
The paper identifies a single direction in the unembedding matrix of large language models that encodes the unigram distribution of the training corpus, acting as a Bayesian prior when the model is uncertain. By projecting the final prediction state onto this direction, the authors derive a per‑token prior loading factor, λ, which decreases as context becomes more informative and decomposes predictions into a tempered prior and a context‑driven likelihood. Experiments across four model families (Llama, Qwen, Gemma, Pythia) show that larger models rely less on the prior in high‑context settings and that manipulating λ can steer predictions toward or away from the unigram prior in KL divergence.