nucleation

LIT-1800: CycleGAN, a Master of Steganography

1 min read

Properties5
StatusActive — read in full 2026-10-29 (NOTE-1346) from arXiv v2; worth reading as the plainest demonstration that a perfect round trip can rest on a channel the intermediate's reader never sees, so that round-trip agreement certifies invertibility, not content. One model, mostly shown by example.
Version2
Tagsrepresentation-learning, information-theory, anthology-candidate
Cited as “paper” byNOTE-1346 — Chu, Zhmoginov and Sandler — CycleGAN as steganographer
Cited as “source” byTHEORY-1127 — A CycleGAN trained to translate aerial photographs to maps and back satisfies its cycle-consistency loss by encoding details of the photograph in a low-amplitude, high-frequency signal in the generated map: the reconstruction recovers details the visible map lacks, is destroyed by noise of three 8-bit levels or by JPEG compression and survives low-frequency change, and the same sensitivity lets an imperceptible perturbation of almost any map make the reverse generator output any chosen photograph, THEORY-1143 — When one domain carries more information than the other, a translation trained only to make the round trip close cannot meet that requirement through a correspondence in the intermediate's visible code, so the round trip closes through a channel the intermediate's audience does not read; agreement of the round trip with its source therefore says nothing about whether the intermediate renders the source's content for that audience

Graph View

Chu, C., Zhmoginov, A., & Sandler, M. (2017). CycleGAN, a master of steganography. NIPS 2017 Workshop on Machine Deception.

Why this is here

A5 called it “the most revealing conceptual warning”: a round trip can be perfect while the intended correspondence is not learned, because the information travels in a channel the audience cannot see. It is the converse failure to CLAIM-496’s collapse, where routes agree because nothing travels, and a learned instance of THEORY-1088’s consensus by changed maps.

The reading (NOTE-1346) bears the description out. One addition: the loop is not wholly unchecked, since discriminators score the map; but they score whether it looks like a map, not whether it says what the photograph shows.

Backlinks