Natural language processing is a hot topic among AI researchers. It includes various tasks like
text classification, sentiment analysis, named entity recognition, machine translation, etc.
Various models have been proposed using both traditional machine learning models and deep
learning models. Models such as OpenAI GPT, ULMFiT, MT-DNN, BERT, etc., achieve state-
of-art results. Fine-tuning the existing models for further improvement requires both extensive
research and domain knowledge. We try to fine-tune the existing models by adopting the
following strategies
- Preprocessing the long text
- Best layer selection
- Selection of appropriate optimizer and learning rate
- Minimizing error rate percentage
- Hyperparameter tuning
The development of new language models based on the existing models is also part of the
research. The new model mainly focuses on the following parameters:
- Proposing new feature extraction techniques
- Word, sentence, paragraph, and document level embeddings
- Focus on the overfitting problem.
felvehető hallgatók száma: 1
Jelentkezési határidő: 2021-08-01
2024. IV. 17. ODT ülés Az ODT következő ülésére 2024. június 14-én, pénteken 10.00 órakor kerül sor a Semmelweis Egyetem Szenátusi termében (Bp. Üllői út 26. I. emelet).
Bejelentkezés
