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

The autoregressive language model (ALM) trained with maximum likelihoodestimation (MLE) is widely used in unconditional text generation. Due toexposure bias, the generated texts still suffer from low quality and diversity.This presents statistically as a discrepancy between the real text andgenerated text. Some research shows a discriminator can detect thisdiscrepancy. Because the discriminator can encode more information than thegenerator, discriminator has the potentiality to improve generator. Toalleviate the exposure bias, generative adversarial networks (GAN) use thediscriminator to update the generator's parameters directly, but they fail bybeing evaluated precisely. A critical reason for the failure is the differencebetween the discriminator input and the ALM input. We propose a novel mechanismby adding a filter which has the same input as the discriminator. First,discriminator detects the discrepancy signals and passes to filter directly (orby learning). Then, we use the filter to reject some generated samples with asampling-based method. Thus, the original generative distribution is revised toreduce the discrepancy. Two ALMs, RNN-based and Transformer-based, areexperimented. Evaluated precisely by three metrics, our mechanism consistentlyoutperforms the ALMs and all kinds of GANs across two benchmark data sets.

Adding A Filter Based on The Discriminator to Improve Unconditional Text Generation