Report Transfer Learning of Deep Convolutional Network on Twitter - Archive ouverte HAL Access content directly
Reports (Research Report) Year : 2017

Report Transfer Learning of Deep Convolutional Network on Twitter

Abstract

This report aims at showing the capacity of transfering a deep neural network on char-level on massive dataset Twitter, using distant supervision. We showed that more data could help for the Stanford140 dataset. The best overal result observed is 84% of transfer learning for two sentiment polarity classes (positive-negative) from 16M emoticons subjective SESAMm dataset to small SemEval 2013 dataset. Other learning of three classes (with neutral) or nine classes based on emoticons (happy, laughing, kisswink, playful, sad, horror, shock, annoyed, hesitated) didn't show any advantages yet in the study.
Fichier principal
Vignette du fichier
Rapport-Transfer-Learning.pdf (745.62 Ko) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-01562179 , version 1 (13-07-2017)

Identifiers

  • HAL Id : hal-01562179 , version 1

Cite

Hoa T Le, Christophe Cerisara, Alexandre Denis. Report Transfer Learning of Deep Convolutional Network on Twitter. [Research Report] Loria & Inria Grand Est. 2017. ⟨hal-01562179⟩
614 View
257 Download

Share

Gmail Mastodon Facebook X LinkedIn More