A pipeline of back-translation, DeBERTa with a learned positional gating scalar, and an attention MLP reports 99.78% accuracy on the Symptom2disease benchmark.
This process involves translating text from English to German and subsequently re-translating it back to English, thereby generating paraphrased versions of the original data
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
An Advanced NLP Framework for Automated Medical Diagnosis with DeBERTa and Dynamic Contextual Positional Gating
A pipeline of back-translation, DeBERTa with a learned positional gating scalar, and an attention MLP reports 99.78% accuracy on the Symptom2disease benchmark.