All sources cited across Survey Meta-Analysis, Expert Synthesis, Quote Collection, Heatmap, and Timeline sections. Every entry links directly to the primary document.
Bell System Technical Journal, Vol. 27, pp. 379–423, 1948
The seminal paper establishing information theory, entropy, and the fundamental limits of data compression — the bedrock on which MDL is built.
Read PDF (Harvard)Bell System Technical Journal, 27: 379–423 · doi:10.1002/j.1538-7305.1948.tb01338.x
Official DOI link to the original 1948 Bell System Technical Journal publication of Shannon's foundational paper.
View at WileyAutomatica, 14(5):465–471, 1978 · ACM DL
The original 1978 paper introducing MDL as a formal principle — the founding text of the entire field.
View at ACM DLAutomatica, Vol. 14, Issue 5, 1978 — ScienceDirect
Full-text access to Rissanen's original Automatica 1978 paper via ScienceDirect.
View at ScienceDirectAnnals of Statistics, Vol. 14, No. 3, pp. 1080–1100, 1986
Introduces the notion of stochastic complexity as the formalization of MDL with optimal universal coding foundations.
Read PDFAnnals of Statistics, 1986 — JSTOR stable link
JSTOR stable entry for Rissanen's stochastic complexity paper — useful for academic citation tracing.
View at JSTORIEEE Transactions on Information Theory, 1998 · doi:10.1109/18.661521
Derives a two-part code version of stochastic complexity showing Fisher-information-optimal quantization; connects MDL to robust regression.
View on Academia.eduProc. 3rd Int. Symposium on Foundations of Quantum Mechanics, Tokyo, 1989
Wheeler's 'it from bit' conjecture: every physical quantity derives from binary yes-or-no observations — a key philosophical foundation of the HS(p)/Mendozian Program.
View at History of InformationPlus Maths — Millennium Mathematics Project, Cambridge
Clear overview of Wheeler's 'it from bit' concept and its significance for quantum mechanics and information theory.
Read explainerMIT Press, 2007 · ISBN 978-0-262-07298-8
The definitive textbook on MDL — comprehensive coverage of two-part codes, NML, stochastic complexity, and applications across statistics, ML, and data mining.
View at MIT PressMIT Press, 2007 — ACM Digital Library entry
ACM Digital Library record for Grünwald's 2007 MDL monograph.
View at ACM DLMIT Press / Neural Information Processing, 2005
Edited volume covering MDL applications in data mining, machine learning, cognitive science, and econometrics.
Read PDFarXiv:math/0406077 — MIT Press chapter preprint
Freely available comprehensive tutorial on MDL — widely cited introduction covering crude MDL, NML, universal codes, and Bayesian connections.
Read on arXivIDSIA Technical Report — PDF
Covers Kolmogorov complexity, Solomonoff induction, and their relationship to MDL and the universal prior.
Read PDFCambridge University Press, 2009 · ISBN 9780521864671
Introduces singular learning theory showing MDL's stochastic complexity behaves differently for singular models (neural nets, HMMs, mixture models).
View at Cambridge UPCambridge Monographs on Applied and Computational Mathematics, 2009
Google Books entry with preview of Watanabe's singular learning theory monograph.
Preview on Google BooksSpringer, excerpted in Routledge chapter PDF
Covers algorithmic probability, universal codes, and MDL — the foundational mathematical text for complexity-based inference.
Read PDF excerptarXiv:2505.13398 · 2025
Uses MDL as a criterion for neural architecture selection — achieves state-of-art compression vs. baseline methods across benchmarks.
Read on arXivarXiv:2007.14009 · 2020
Applies MDL-based two-part codes to detect anomalies and structural breaks in multivariate time series with better precision than statistical baselines.
Read on arXivChaos: An Interdisciplinary Journal of Nonlinear Science, Vol. 35, 043132 · 2025
Uses MDL to select sparse readout subsets in echo-state networks — improves prediction of chaotic systems (Lorenz, Rössler, Thomas) vs. ridge regression.
Read at AIPAdvances in Neural Information Processing Systems 36 (NeurIPS 2023)
Demonstrates MDL-based representation learning achieves superior generalization and interpretability vs. cross-entropy baselines on benchmark datasets.
Read at NeurIPSProceedings of ACL 2024, Long Papers
Proposes an MDL-inspired IBE-Eval framework to assess the quality of LLM-generated explanations — measures parsimony and coherence of reasoning chains.
Read at ACL AnthologyAdvances in Neural Information Processing Systems 30 (NIPS 2017)
Shows that Bayesian sparsity-inducing priors implement MDL-style compression in neural networks — achieves state-of-art compression on LeNet and VGG architectures.
Read PDF (NIPS)NeurIPS 2017 — ACM Digital Library
ACM Digital Library entry for the Louizos et al. Bayesian compression paper.
View at ACM DLMDL+DM Workshop, KDD 2019 — CISPA
Tutorial slides covering modern MDL for data mining, including the Krimp algorithm and pattern selection by minimum description length.
Read PDF (CISPA)CISPA Publications, 2021
Comprehensive tutorial on modern MDL for data mining with theory and practical guidance on model selection and pattern mining.
View at CISPACWI Technical Report, 2004 — PDF mirror
Earlier 2004 tutorial version (precursor to the 2007 book) — more accessible entry point into MDL theory.
Read PDFACL 2024 Long Papers, pp. 2462–2475, Bangkok
Introduces Reasoning with Attributions to improve multi-hop reasoning in LMs — related to MDL parsimony in chain-of-thought evaluation.
Read at ACL AnthologyACL 2024 Long Papers
Proposes structured knowledge triple reasoning chains to improve LLM faithfulness — related to MDL-optimal explanation compression.
Read at ACL AnthologyMorgan Kaufmann — PDF excerpt
Foundational decision tree learning — uses information-theoretic (MDL-adjacent) criteria for tree pruning and model selection.
Read PDFYouTube · Udacity
Short lecture on MDL fundamentals: two-part codes, model complexity, and application to supervised learning. Widely used in ML courses.
Watch on YouTubeYouTube · IIT Madras
University lecture covering MDL from first principles: Kolmogorov complexity, Shannon entropy, and the formal derivation of the two-part code.
Watch on YouTubeYouTube · IIT Madras
Continuation covering NML, stochastic complexity, and practical MDL model selection procedures.
Watch on YouTubePIBBSS (Principles of Intelligent Behavior in Biological and Social Systems) 2024
Explores connections between MDL, Friston's Free Energy Principle, and active inference — directly relevant to the H² Framework's cross-domain unification.
View PIBBSS 2024YouTube
Expert discussion of how MDL applies to neural network model selection and its relationship to regularization techniques.
Watch on YouTubeYouTube · 2024
Discussion of MDL applied to robust model selection and out-of-distribution generalization in modern AI systems.
Watch on YouTubeYouTube · 2023
Traces MDL's roots in information theory — covers Shannon, Kolmogorov, and Rissanen in accessible form for AI practitioners.
Watch on YouTubeEnglish Wikipedia
Comprehensive encyclopedic overview of MDL including history, formulations, and applications — good for quick cross-referencing.
Read on WikipediaEnglish Wikipedia
Biography of Jorma Rissanen — the Finnish mathematician and information theorist who created the MDL principle in 1978.
Read on WikipediaEnglish Wikipedia
Overview of Shannon's 1948 paper including historical context, key theorems, and long-term impact on compression and MDL.
Read on WikipediaTokyo Institute of Technology / Google Sites
Watanabe's official research homepage explaining singular learning theory and its relationship to MDL stochastic complexity in deep learning.
Visit homepageemergentmind.com
AI-curated topic hub aggregating recent arXiv papers on MDL — useful for tracking current literature trends.
Visit hubpibbss.ai
Official page for the 2024 PIBBSS Symposium on intelligent behavior in biological and social systems — features MDL/FEP cross-domain talks.
Visit PIBBSS 2024ITSOC Newsletter, Sept 2020
Official IEEE obituary for Rissanen — includes quotes and assessment of his contributions to MDL and information theory.
Read PDF (IEEE)CWI Amsterdam — ir.cwi.nl
Full PDF of the MIT Press MDL monograph hosted by CWI Amsterdam — the most accessible complete source.
Read PDF (CWI)