English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Over-parameterized neural language models (LMs) can memorize and recite longsequences of training data. While such memorization is normally associated withundesired properties such as overfitting and information leaking, our workcasts memorization as an unexplored capability of LMs. We propose the firstsymmetric encryption algorithm with autoregressive language models (SELM). Weshow that autoregressive LMs can encode arbitrary data into a compactreal-valued vector (i.e., encryption) and then losslessly decode the vector tothe original message (i.e., decryption) via random subspace optimization andgreedy decoding. While SELM is not amenable to conventional cryptanalysis, weinvestigate its security through a novel empirical variant of the classicIND-CPA (indistinguishability under chosen-plaintext attack) game and showpromising results on security. Our code and datasets are available athttps://github.com/OSU-NLP-Group/SELM.

Memorization for Good: Encryption with Autoregressive Language Models