Information Theory

Also known as: Shannon Theory

framework · mathematics · formal-scientific

Shannon's framework quantifying information, entropy, and communication channels.

Information Theory was founded by Claude Shannon in his 1948 Bell System Technical Journal paper 'A Mathematical Theory of Communication' (substantially expanded into book form by Shannon and Warren Weaver in 1949), creating an entirely new mathematical discipline that quantifies information independent of meaning or content. Shannon's foundational results include: information measured in bits as the logarithm of the number of equally likely possibilities; entropy as the average information content of a probability distribution (H = -Σp log p); the source coding theorem (data can be compressed to its entropy but no further); the noisy channel coding theorem (reliable communication is possible at any rate below the channel capacity); mutual information as the reduction in uncertainty of one variable given another. Subsequent development has been enormous: error-correcting codes (Hamming, Reed-Solomon, LDPC, Turbo codes); Kullback-Leibler divergence; algorithmic information theory (Kolmogorov, Chaitin, Solomonoff); rate-distortion theory; network information theory; quantum information theory. The framework has had outsized impact on telecommunications (the entire digital-communication infrastructure rests on Shannon's results), data compression (ZIP, JPEG, MP3, MPEG), cryptography (Shannon's 1949 'Communication Theory of Secrecy Systems' founded modern cryptography), machine learning (cross-entropy, KL divergence, mutual-information-based feature selection), neuroscience (information-theoretic analysis of neural codes), and statistical physics (the connection between Shannon entropy and thermodynamic entropy). Shannon's framework is one of the most generative single-author intellectual achievements of the 20th century.

Originators

Claude E. Shannon (foundational); intellectual antecedents in Harry Nyquist, Ralph Hartley; Warren Weaver (substantial coauthor of 1949 book) high

Year / Decade

1948 (Shannon paper); 1949 (Shannon-Weaver book); ongoing development high

Primary sources

Shannon, C.E. (1948). 'A Mathematical Theory of Communication', Bell System Technical Journal, Shannon, C.E. & Weaver, W. (1949). The Mathematical Theory of Communication, Shannon, C.E. (1949). 'Communication Theory of Secrecy Systems', Bell System Technical Journal, Cover, T.M. & Thomas, J.A. (2006, 2nd ed.). Elements of Information Theory high

Core components

Primary use case

Foundation of digital communication theory (the entire modern telecommunications infrastructure); data compression algorithms and standards; cryptography (Shannon's 1949 secrecy paper as founding work); machine learning (cross-entropy loss, KL divergence, information-theoretic regularization); neuroscience and computational biology (information in neural codes, genetic codes); statistical physics (Shannon entropy and thermodynamic entropy); statistical inference (MDL, AIC, mutual-information-based methods); pedagogical reference in electrical engineering, computer science, and mathematics curricula.

Common criticisms

Lineage