LSTM sequence generation with mixture-density outputs and soft alignment

What changed

Demonstrated recurrent text/continuous handwriting generation and conditioned handwriting synthesis on text through a learned, forward-moving Gaussian-mixture soft window.

Scientific scope

Demonstrated recurrent text/continuous handwriting generation and conditioned handwriting synthesis on text through a learned, forward-moving Gaussian-mixture soft window. Extends existing RNN/LSTM and mixture-density methods, not invention of all autoregressive generation or later Transformer attention. Source soft-window weights are unnormalized and not literal character probabilities; sampled style realism does not certify transcription fidelity, arbitrary long-range coherence or authorship authenticity. Original2013single-author report only.

Evidence

The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Extends existing RNN/LSTM and mixture-density methods, not invention of all autoregressive generation or later Transformer attention. Source soft-window weights are unnormalized and not literal character probabilities; sampled style realism does not certify transcription fidelity, arbitrary long-range coherence or authorship authenticity. Original2013single-author report only.

Contributors

• Source-listed authors and technical contributors · SCIENCE: Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.

• Generating Sequences With Recurrent Neural Networks

Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.

Full scientific record · Browse contribution grades · Request a correction