Loading...
| Friend's email | |
| Your name | |
| Your email | |
| enter code | |
This page was sent successfuly
- Type of Document: M.Sc. Thesis
- Language: Farsi
- Document No: 58492 (05)
- University: Sharif University of Technology
- Department: Electrical Engineering
- Advisor(s): Bagheri Shouraki, Saeed
- Abstract:
- One of the distinctive features of human intelligence is the ability to quickly learn new concepts and generalize knowledge to novel situations, whereas machine learning models typically require large volumes of data to achieve optimal performance. Meta-learning, a subfield of machine learning, aims to train models capable of rapidly adapting to new tasks using only a few training examples. This thesis explores a relatively underexplored area by leveraging the capabilities of meta-learning to classify texts generated either by humans or by generative language models—an increasingly significant challenge in natural language processing due to the proliferation of large language models and the scarcity of labeled data. A hybrid model is proposed, which combines linguistic and structural features derived from a human experiment with semantic representations produced by a pre-trained language model. Using a meta-learning-based algorithm, the model classifies human- and machine-generated texts under conditions of limited training data. Trained through diverse and small-scale tasks, the proposed model demonstrates superior accuracy and generalizability compared to traditional methods. This research represents a novel step toward applying meta-learning to address an emerging issue in the field of natural language processing
- Keywords:
- Metalearning ; Text Classification ; Large Language Model ; Natural Language Processing ; Human Visual System ; Human vs. Machine Text Classification
Digital Object List
-
محتواي کتاب
- view
Bookmark