Active inference and artificial reasoning
- URL: http://arxiv.org/abs/2512.21129v1
- Date: Wed, 24 Dec 2025 11:59:36 GMT
- Title: Active inference and artificial reasoning
- Authors: Karl Friston, Lancelot Da Costa, Alexander Tschantz, Conor Heins, Christopher Buckley, Tim Verbelen, Thomas Parr,
- Abstract summary: This technical note considers the sampling of outcomes that provide the greatest amount of information about the structure of underlying world models.<n>We focus on the sample efficiency afforded by seeking outcomes that resolve the greatest uncertainty about the world model.
- Score: 36.949648744325046
- License: http://creativecommons.org/licenses/by/4.0/
- Abstract: This technical note considers the sampling of outcomes that provide the greatest amount of information about the structure of underlying world models. This generalisation furnishes a principled approach to structure learning under a plausible set of generative models or hypotheses. In active inference, policies - i.e., combinations of actions - are selected based on their expected free energy, which comprises expected information gain and value. Information gain corresponds to the KL divergence between predictive posteriors with, and without, the consequences of action. Posteriors over models can be evaluated quickly and efficiently using Bayesian Model Reduction, based upon accumulated posterior beliefs about model parameters. The ensuing information gain can then be used to select actions that disambiguate among alternative models, in the spirit of optimal experimental design. We illustrate this kind of active selection or reasoning using partially observed discrete models; namely, a 'three-ball' paradigm used previously to describe artificial insight and 'aha moments' via (synthetic) introspection or sleep. We focus on the sample efficiency afforded by seeking outcomes that resolve the greatest uncertainty about the world model, under which outcomes are generated.
Related papers
- Modeling and Discovering Direct Causes for Predictive Models [0.0]
We introduce a causal modeling framework that captures the input-output behavior of predictive models.<n>We then present sound and complete algorithms for discovering direct causes (from data) under some assumptions.
arXiv Detail & Related papers (2024-12-03T22:25:42Z) - Fast Explanations via Policy Gradient-Optimized Explainer [7.011763596804071]
This paper introduces a novel framework that represents attribution-based explanations via probability distributions.<n>The proposed framework offers a robust, scalable solution for real-time, large-scale model explanations.<n>We validate our framework on image and text classification tasks and the experiments demonstrate that our method reduces inference time by over 97% and memory usage by 70%.
arXiv Detail & Related papers (2024-05-29T00:01:40Z) - A PAC-Bayesian Perspective on the Interpolating Information Criterion [54.548058449535155]
We show how a PAC-Bayes bound is obtained for a general class of models, characterizing factors which influence performance in the interpolating regime.
We quantify how the test error for overparameterized models achieving effectively zero training error depends on the quality of the implicit regularization imposed by e.g. the combination of model, parameter-initialization scheme.
arXiv Detail & Related papers (2023-11-13T01:48:08Z) - Consistent Explanations in the Face of Model Indeterminacy via
Ensembling [12.661530681518899]
This work addresses the challenge of providing consistent explanations for predictive models in the presence of model indeterminacy.
We introduce ensemble methods to enhance the consistency of the explanations provided in these scenarios.
Our findings highlight the importance of considering model indeterminacy when interpreting explanations.
arXiv Detail & Related papers (2023-06-09T18:45:43Z) - Control-Oriented Model-Based Reinforcement Learning with Implicit
Differentiation [11.219641045667055]
We propose an end-to-end approach for model learning which directly optimize the expected returns using implicit differentiation.
We provide theoretical and empirical evidence highlighting the benefits of our approach in the model misspecification regime compared to likelihood-based methods.
arXiv Detail & Related papers (2021-06-06T23:15:49Z) - Attentional Prototype Inference for Few-Shot Segmentation [128.45753577331422]
We propose attentional prototype inference (API), a probabilistic latent variable framework for few-shot segmentation.
We define a global latent variable to represent the prototype of each object category, which we model as a probabilistic distribution.
We conduct extensive experiments on four benchmarks, where our proposal obtains at least competitive and often better performance than state-of-the-art prototype-based methods.
arXiv Detail & Related papers (2021-05-14T06:58:44Z) - Beyond Trivial Counterfactual Explanations with Diverse Valuable
Explanations [64.85696493596821]
In computer vision applications, generative counterfactual methods indicate how to perturb a model's input to change its prediction.
We propose a counterfactual method that learns a perturbation in a disentangled latent space that is constrained using a diversity-enforcing loss.
Our model improves the success rate of producing high-quality valuable explanations when compared to previous state-of-the-art methods.
arXiv Detail & Related papers (2021-03-18T12:57:34Z) - Characterizing Fairness Over the Set of Good Models Under Selective
Labels [69.64662540443162]
We develop a framework for characterizing predictive fairness properties over the set of models that deliver similar overall performance.
We provide tractable algorithms to compute the range of attainable group-level predictive disparities.
We extend our framework to address the empirically relevant challenge of selectively labelled data.
arXiv Detail & Related papers (2021-01-02T02:11:37Z) - Control as Hybrid Inference [62.997667081978825]
We present an implementation of CHI which naturally mediates the balance between iterative and amortised inference.
We verify the scalability of our algorithm on a continuous control benchmark, demonstrating that it outperforms strong model-free and model-based baselines.
arXiv Detail & Related papers (2020-07-11T19:44:09Z) - Learning Opinion Dynamics From Social Traces [25.161493874783584]
We propose an inference mechanism for fitting a generative, agent-like model of opinion dynamics to real-world social traces.
We showcase our proposal by translating a classical agent-based model of opinion dynamics into its generative counterpart.
We apply our model to real-world data from Reddit to explore the long-standing question about the impact of backfire effect.
arXiv Detail & Related papers (2020-06-02T14:48:17Z) - Capsule Networks -- A Probabilistic Perspective [42.187785678596384]
'Capsule' models try to explicitly represent the poses of objects, enforcing a linear relationship between an object's pose and that of its constituent parts.
We describe a probabilistic generative model which encodes such capsule assumptions, clearly separating the generative parts of the model from the inference mechanisms.
We experimentally demonstrate the applicability of our unified objective, and demonstrate the use of test time optimisation to solve problems inherent to amortised inference in our model.
arXiv Detail & Related papers (2020-04-07T17:26:11Z) - Decision-Making with Auto-Encoding Variational Bayes [71.44735417472043]
We show that a posterior approximation distinct from the variational distribution should be used for making decisions.
Motivated by these theoretical results, we propose learning several approximate proposals for the best model.
In addition to toy examples, we present a full-fledged case study of single-cell RNA sequencing.
arXiv Detail & Related papers (2020-02-17T19:23:36Z)
This list is automatically generated from the titles and abstracts of the papers in this site.
This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.