Matches in SemOpenAlex for { <https://semopenalex.org/work/W3136786651> ?p ?o ?g. }
- W3136786651 abstract "Adversarial training of end-to-end (E2E) ASR systems using generative adversarial networks (GAN) has recently been explored for low-resource ASR corpora. GANs help to learn the true data representation through a two-player min-max game. However, training an E2E ASR model using a large ASR corpus with a GAN framework has never been explored, because it might take excessively long time due to high-variance gradient updates and face convergence issues. In this paper, we introduce a novel framework for fine-tuning a pre-trained ASR model using the GAN objective where the ASR model acts as a generator and a discriminator tries to distinguish the ASR output from the real data. Since the ASR model is pre-trained, we hypothesize that the ASR model output (soft distribution vectors) helps to get higher scores from the discriminator and makes the task of the discriminator harder within our GAN framework, which in turn improves the performance of the ASR model in the fine-tuning stage. Here, the pre-trained ASR model is fine-tuned adversarially against the discriminator using an additional adversarial loss. Experiments on full LibriSpeech dataset show that our proposed approach outperforms baselines and conventional GAN-based adversarial models." @default.
- W3136786651 created "2021-03-29" @default.
- W3136786651 creator A5006199508 @default.
- W3136786651 creator A5028862918 @default.
- W3136786651 date "2021-03-10" @default.
- W3136786651 modified "2023-09-27" @default.
- W3136786651 title "Fine-tuning of Pre-trained End-to-end Speech Recognition with Generative Adversarial Networks" @default.
- W3136786651 cites W1494198834 @default.
- W3136786651 cites W1828163288 @default.
- W3136786651 cites W1895481600 @default.
- W3136786651 cites W2099471712 @default.
- W3136786651 cites W2102113734 @default.
- W3136786651 cites W2127141656 @default.
- W3136786651 cites W2327501763 @default.
- W3136786651 cites W2514741789 @default.
- W3136786651 cites W2605135824 @default.
- W3136786651 cites W2739748921 @default.
- W3136786651 cites W2766210095 @default.
- W3136786651 cites W2795050058 @default.
- W3136786651 cites W2795935804 @default.
- W3136786651 cites W2885185669 @default.
- W3136786651 cites W2892356933 @default.
- W3136786651 cites W2913851961 @default.
- W3136786651 cites W2936774411 @default.
- W3136786651 cites W2962760690 @default.
- W3136786651 cites W2963341071 @default.
- W3136786651 cites W2963362078 @default.
- W3136786651 cites W2963400424 @default.
- W3136786651 cites W2963403868 @default.
- W3136786651 cites W2964201867 @default.
- W3136786651 cites W2972389417 @default.
- W3136786651 cites W2972818416 @default.
- W3136786651 cites W2981857663 @default.
- W3136786651 cites W2990205979 @default.
- W3136786651 cites W2991213871 @default.
- W3136786651 cites W2992632249 @default.
- W3136786651 cites W2998814410 @default.
- W3136786651 cites W3005078977 @default.
- W3136786651 cites W3015974384 @default.
- W3136786651 cites W3097649098 @default.
- W3136786651 cites W385466589 @default.
- W3136786651 hasPublicationYear "2021" @default.
- W3136786651 type Work @default.
- W3136786651 sameAs 3136786651 @default.
- W3136786651 citedByCount "0" @default.
- W3136786651 crossrefType "posted-content" @default.
- W3136786651 hasAuthorship W3136786651A5006199508 @default.
- W3136786651 hasAuthorship W3136786651A5028862918 @default.
- W3136786651 hasConcept C121332964 @default.
- W3136786651 hasConcept C137293760 @default.
- W3136786651 hasConcept C154945302 @default.
- W3136786651 hasConcept C162324750 @default.
- W3136786651 hasConcept C163258240 @default.
- W3136786651 hasConcept C2777303404 @default.
- W3136786651 hasConcept C2779803651 @default.
- W3136786651 hasConcept C2780992000 @default.
- W3136786651 hasConcept C28490314 @default.
- W3136786651 hasConcept C37736160 @default.
- W3136786651 hasConcept C39890363 @default.
- W3136786651 hasConcept C41008148 @default.
- W3136786651 hasConcept C50522688 @default.
- W3136786651 hasConcept C62520636 @default.
- W3136786651 hasConcept C74296488 @default.
- W3136786651 hasConcept C76155785 @default.
- W3136786651 hasConcept C94915269 @default.
- W3136786651 hasConceptScore W3136786651C121332964 @default.
- W3136786651 hasConceptScore W3136786651C137293760 @default.
- W3136786651 hasConceptScore W3136786651C154945302 @default.
- W3136786651 hasConceptScore W3136786651C162324750 @default.
- W3136786651 hasConceptScore W3136786651C163258240 @default.
- W3136786651 hasConceptScore W3136786651C2777303404 @default.
- W3136786651 hasConceptScore W3136786651C2779803651 @default.
- W3136786651 hasConceptScore W3136786651C2780992000 @default.
- W3136786651 hasConceptScore W3136786651C28490314 @default.
- W3136786651 hasConceptScore W3136786651C37736160 @default.
- W3136786651 hasConceptScore W3136786651C39890363 @default.
- W3136786651 hasConceptScore W3136786651C41008148 @default.
- W3136786651 hasConceptScore W3136786651C50522688 @default.
- W3136786651 hasConceptScore W3136786651C62520636 @default.
- W3136786651 hasConceptScore W3136786651C74296488 @default.
- W3136786651 hasConceptScore W3136786651C76155785 @default.
- W3136786651 hasConceptScore W3136786651C94915269 @default.
- W3136786651 hasLocation W31367866511 @default.
- W3136786651 hasOpenAccess W3136786651 @default.
- W3136786651 hasPrimaryLocation W31367866511 @default.
- W3136786651 hasRelatedWork W1504449689 @default.
- W3136786651 hasRelatedWork W2804884650 @default.
- W3136786651 hasRelatedWork W2903015118 @default.
- W3136786651 hasRelatedWork W2913566369 @default.
- W3136786651 hasRelatedWork W2952010730 @default.
- W3136786651 hasRelatedWork W2992448548 @default.
- W3136786651 hasRelatedWork W3081494755 @default.
- W3136786651 hasRelatedWork W3094225009 @default.
- W3136786651 hasRelatedWork W3095773170 @default.
- W3136786651 hasRelatedWork W3097573669 @default.
- W3136786651 hasRelatedWork W3105804405 @default.
- W3136786651 hasRelatedWork W3131305672 @default.
- W3136786651 hasRelatedWork W3137489363 @default.
- W3136786651 hasRelatedWork W3147607624 @default.
- W3136786651 hasRelatedWork W3155168732 @default.