GENERATIVE OF ORIGIN MODEL DISTRIBUTION MASKED WITH EMOTIONS AND TOPICS DISTRIBUTION IN HYBRID METHOD

Woohyun Park, Dongryeol Shin

Keywords: Masked Distribution, Learning Approximation Representation, Topic Analytics, Sentiment Analytics, Meta Learning

TL;DR: Embedding constructors for effective representation of natural language

Abstract: There is a vast amount of data in the form of natural signals in the world, and difficult expression processors are required to analyze such data. Traditional embedding methods are susceptible to generalization failure. In this study, we developed a classification model that creates and approximates an origin hypothesis model using limited emotions and topics. To solve the hypothesis, the proposed model utilizes dynamic learner modules. Using this mechanism, a text-based origin distribution representation learning model was designed. In order to simulate and generalize, we analyzed the experimental evaluation results via various natural language data sets and measured the corresponding performance. Thus, we demonstrated that the machine achieves the classification task more effectively by integrating learning distribution and multiple learning methods.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

Submission Guidelines: Yes

Please Choose The Closest Area That Your Submission Falls Into: Deep Learning and representational learning