On-device research index

arXiv research

A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.

169,181 papers · 148 categories

Trend · papers per month

count % of corpus

3 results for “SCST”

B-SCST: Bayesian Self-Critical Sequence Training for Image Captioningcs.LG

Bayesian approach improves image captioning quality metrics.

problem Improving image captioning quality metrics.

method Bayesian Self-Critical Sequence Training (B-SCST) with Monte Carlo dropout approximate variational inference.

result B-SCST improves CIDEr-D scores on various image captioning datasets.

image captioning Bayesian deep learning Monte Carlo dropout CIDEr-D reinforcement learning

2020-04-06abs ↗pdf ↗

Adversarial Semantic Alignment for Improved Image Captionscs.LG

Improved image captions through adversarial semantic alignment.

problem Automatic evaluation and generalization to unseen compositions.

method Context-aware LSTM captioner and co-attentive discriminator for semantic alignment; SCST training method.

result SCST training method shows better performance in semantic score and human evaluation.

image captioning conditional GAN SCST semantic alignment automatic evaluation

2018-04-30abs ↗pdf ↗

A Reinforced Topic-Aware Convolutional Sequence-to-Sequence Model for Abstractive Text Summarizationcs.CL

A new model improves text summarization by integrating topic information.

problem Improving the coherence, diversity, and informativeness of text summarization.

method Integrates topic information into ConvS2S model and uses SCST for optimization.

result The proposed model outperforms state-of-the-art methods in abstractive summarization.

text summarization topic-aware convolutional sequence-to-sequence reinforcement learning SCST

2018-05-09abs ↗pdf ↗