Date of Award
10-29-2019
Document Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
Department
Computer Science
First Advisor
Xiaowei Xu
Abstract
[Motivation] Machine learning and deep learning models is the main methodology for discovering hidden knowledge of data. Data plays a significant role in the training process. However, lack of data especially labeled data is precious due to accessibility and high cost. How to use the abundant unlabeled data to enhance the downstream learning process on labeled data is an urgent task in machine learning and pattern recognition. [Methods] An STL (Self-taught learning) model is proposed in this dissertation, which can successfully learn features from downloaded unlabeled text and enhance the downstream tasks on labeled text. In addition, DLI-IT (Drug labeling identification through image and text) system is built to retrieve information based on image and transferred text features from Google Sentence embedding. Additionally, the IT-STL (Iterative Self-taught learning) model is created to iteratively transfer features from unlabeled text based on Generative Adversarial Neural Network, which can make a better enhancement in an efficient way. [Results] STL model successfully transferred features from unlabeled data and enhances downstream tasks. DLI-IT model can identify drugs on accuracy up-to 90%. In addition, the IT-STL model iteratively transfer features from unlabeled data and model converge after 4th or 5th iteration on review datasets. It also achieves the state-of-the-art results in an efficient way.
Recommended Citation
Liu, Xiangwen, "Deep Neural Network Based Iterative Self-Taught Learning on Text Mining" (2019). Theses and Dissertations. 898.
https://research.ualr.edu/etd/898
