Open problems in universal induction \& intelligence

DOI10.3390/A2030879zbMATH Open1461.68170arXiv0907.0746OpenAlexW2132450394WikidataQ58012382 ScholiaQ58012382MaRDI QIDQ1662486FDOQ1662486

Authors: Marcus Hutter 

Publication date: 20 August 2018

Published in: Algorithms (Search for Journal in Brave)

Abstract: Specialized intelligent systems can be found everywhere: finger print, handwriting, speech, and face recognition, spam filtering, chess and other game programs, robots, et al. This decade the first presumably complete mathematical theory of artificial intelligence based on universal induction-prediction-decision-action has been proposed. This information-theoretic approach solidifies the foundations of inductive inference and artificial intelligence. Getting the foundations right usually marks a significant progress and maturing of a field. The theory provides a gold standard and guidance for researchers working on intelligent algorithms. The roots of universal induction have been laid exactly half-a-century ago and the roots of universal intelligence exactly one decade ago. So it is timely to take stock of what has been achieved and what remains to be done. Since there are already good recent surveys, I describe the state-of-the-art only in passing and refer the reader to the literature. This article concentrates on the open problems in universal induction and its extension to universal intelligence.

Full work available at URL: https://arxiv.org/abs/0907.0746

Recommendations

• Towards a Universal Theory of Artificial Intelligence based on Algorithmic Probability and Sequential Decision Theory
• Asymptotic non-learnability of universal agents with computable horizon functions
• Universal artificial intelligence. Sequential decisions based on algorithmic probability.
• Is There an Elegant Universal Theory of Prediction?
• Algorithmic Probability: Theory and Applications

zbMATH Keywords

artificial intelligenceinformation theoryKolmogorov complexityreinforcement learningrational agentssequential decision theoryuniversal Solomonoff induction

Mathematics Subject Classification ID

Learning and adaptive systems in artificial intelligence (68T05) Algorithmic information theory (Kolmogorov complexity, etc.) (68Q30) General topics in artificial intelligence (68T01)

Cites Work

• Online planning algorithms for POMDPS
• Introduction to Time Series and Forecasting
• Elements of Information Theory
• Title not available (Why is that?)
• Prediction, Learning, and Games
• Pattern recognition and machine learning.
• Title not available (Why is that?)
• Title not available (Why is that?)
• Planning and acting in partially observable stochastic domains
• Title not available (Why is that?)
• Time-Series Forecasting
• Title not available (Why is that?)
• The Selection of Prior Distributions by Formal Rules
• \({\mathcal Q}\)-learning
• Title not available (Why is that?)
• An invariant form for the prior probability in estimation problems
• Title not available (Why is that?)
• Title not available (Why is that?)
• Probability Theory
• Title not available (Why is that?)
• Title not available (Why is that?)
• Subjective Bayesian analysis: principles and practice
• Modeling by shortest data description
• THE COMPLEXITY OF FINITE OBJECTS AND THE DEVELOPMENT OF THE CONCEPTS OF INFORMATION AND RANDOMNESS BY MEANS OF THE THEORY OF ALGORITHMS
• The definition of random sequences
• Language identification in the limit
• A formal theory of inductive inference. Part II
• Title not available (Why is that?)
• Rational Learning Leads to Nash Equilibrium
• Clustering by Compression
• Title not available (Why is that?)
• Title not available (Why is that?)
• An Information Measure for Classification
• On the Complexity of Finite Sequences
• The Uncertain Reasoner's Companion
• Algorithmic statistics
• On the Length of Programs for Computing Finite Binary Sequences
• On Computable Numbers, with an Application to the Entscheidungsproblem
• An introduction to Kolmogorov complexity and its applications
• Asymptotics of Discrete MDL for Online Prediction
• Randomness conservation inequalities; information and independence in mathematical theories
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Complexity-based induction systems: Comparisons and convergence theorems
• Universal artificial intelligence. Sequential decisions based on algorithmic probability.
• Title not available (Why is that?)
• Title not available (Why is that?)
• Self-Optimizing and Pareto-Optimal Policies in General Environments based on Bayes-Mixtures
• General Discounting Versus Average Reward
• On universal prediction and Bayesian confirmation
• Convergence and loss bounds for bayesian sequence prediction
• 10.1162/1532443041827952
• Algorithmic complexity bounds on future prediction errors
• Gödel's theorem and information
• Information, randomness and incompleteness. Papers on algorithmic information theory
• Sequential predictions based on algorithmic complexity
• HIERARCHIES OF GENERALIZED KOLMOGOROV COMPLEXITIES AND NONENUMERABLE UNIVERSAL MEASURES COMPUTABLE IN THE LIMIT
• Inferring decision trees using the minimum description length principle
• On the possibility of learning in reactive environments with arbitrary dependence
• Title not available (Why is that?)
• Title not available (Why is that?)
• Towards a Universal Theory of Artificial Intelligence based on Algorithmic Probability and Sequential Decision Theory
• Kolmogorov entropy in the context of computability theory
• Theory and Applications of Models of Computation
• Algorithmic Learning Theory
• Optimal ordered problem solver
• Convergence and Error Bounds for Universal Prediction of Nonbinary Sequences
• Title not available (Why is that?)
• Learning Theory
• Sequence prediction based on monotone complexity.
• Algorithmic Learning Theory
• On the existence and convergence of computable universal priors
• New error bounds for Solomonoff prediction
• On generalized computable universal priors and their convergence
• Algorithmic Learning Theory
• Title not available (Why is that?)
• Fast online \(Q(\lambda)\)
• Title not available (Why is that?)
• Predicting non-stationary processes
• Theory and Applications of Models of Computation
• On semimeasures predicting Martin-Löf random sequences
• THE FASTEST AND SHORTEST ALGORITHM FOR ALL WELL-DEFINED PROBLEMS
• Universal forecasting algorithms
• Asymptotically efficient adaptive allocation schemes for controlled Markov chains: finite parameter space
• Asymptotically efficient adaptive allocation schemes for controlled i.i.d. processes: finite parameter space
• Title not available (Why is that?)
• Beyond the Turing test
• Toward a model of intelligence as an economy of agents
• Title not available (Why is that?)
• Asymptotic Learnability of Reinforcement Problems with Arbitrary Dependence
• Title not available (Why is that?)
• Algorithmic Learning Theory
• Title not available (Why is that?)
• Title not available (Why is that?)

Cited In (5)

• A philosophical treatise of universal induction
• Towards a Universal Theory of Artificial Intelligence based on Algorithmic Probability and Sequential Decision Theory
• Induction and Physical Theory Formation As Well As Universal Computation by Machine Learning
• Universal artificial intelligence. Sequential decisions based on algorithmic probability.
• Putnam's diagonal argument and the impossibility of a universal learning machine

Uses Software

• POMDPS
• itsmr
• PRMLT

This page was built for publication: Open problems in universal induction \& intelligence

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1662486)