JUST at VQA-Med: A VGG-Seq2Seq model

Bashar Talafha; Mahmoud Al-Ayyoub

JUST at VQA-Med: A VGG-Seq2Seq model

Bashar Talafha
, Mahmoud Al-Ayyoub

Jordan University of Science and Technology

Research output: Contribution to journal › Conference article › peer-review

3 Scopus citations

Abstract

This paper describes the VGG-Seq2Seq system for the Medical Domain Visual Question Answering (VQA-Med) Task of ImageCLEF 2018. The proposed system follows the encoder-decoder architecture, where the encoders fuses a pretrained VGG network with an LSTM network that has a pretrained word embedding layer to encode the input. To generate the output, another LSTM network is used for decoding. When used with a pretrained VGG network, the VGG-Seq2Seq model managed to achieve reasonable results with 0.06, 0.12, 0.03 BLEU, WBSS and CBSS, respectively. Moreover, the VGG-Seq2Seq is not expensive to train.

Original language	English
Journal	CEUR Workshop Proceedings
Volume	2125
State	Published - 2018
Externally published	Yes
Event	19th Working Notes of CLEF Conference and Labs of the Evaluation Forum, CLEF 2018 - Avignon, France Duration: 10 Sep 2018 → 14 Sep 2018

Keywords

Global Vectors for Word Representation
Sequence to sequence
VGG Network

Cite this

@article{9603fc46d8634091b49203656e76e444,

title = "JUST at VQA-Med: A VGG-Seq2Seq model",

abstract = "This paper describes the VGG-Seq2Seq system for the Medical Domain Visual Question Answering (VQA-Med) Task of ImageCLEF 2018. The proposed system follows the encoder-decoder architecture, where the encoders fuses a pretrained VGG network with an LSTM network that has a pretrained word embedding layer to encode the input. To generate the output, another LSTM network is used for decoding. When used with a pretrained VGG network, the VGG-Seq2Seq model managed to achieve reasonable results with 0.06, 0.12, 0.03 BLEU, WBSS and CBSS, respectively. Moreover, the VGG-Seq2Seq is not expensive to train.",

keywords = "Global Vectors for Word Representation, Sequence to sequence, VGG Network",

author = "Bashar Talafha and Mahmoud Al-Ayyoub",

year = "2018",

language = "English",

volume = "2125",

journal = "CEUR Workshop Proceedings",

issn = "1613-0073",

publisher = "CEUR-WS",

note = "19th Working Notes of CLEF Conference and Labs of the Evaluation Forum, CLEF 2018 ; Conference date: 10-09-2018 Through 14-09-2018",

}

TY - JOUR

T1 - JUST at VQA-Med

T2 - 19th Working Notes of CLEF Conference and Labs of the Evaluation Forum, CLEF 2018

AU - Talafha, Bashar

AU - Al-Ayyoub, Mahmoud

PY - 2018

Y1 - 2018

N2 - This paper describes the VGG-Seq2Seq system for the Medical Domain Visual Question Answering (VQA-Med) Task of ImageCLEF 2018. The proposed system follows the encoder-decoder architecture, where the encoders fuses a pretrained VGG network with an LSTM network that has a pretrained word embedding layer to encode the input. To generate the output, another LSTM network is used for decoding. When used with a pretrained VGG network, the VGG-Seq2Seq model managed to achieve reasonable results with 0.06, 0.12, 0.03 BLEU, WBSS and CBSS, respectively. Moreover, the VGG-Seq2Seq is not expensive to train.

AB - This paper describes the VGG-Seq2Seq system for the Medical Domain Visual Question Answering (VQA-Med) Task of ImageCLEF 2018. The proposed system follows the encoder-decoder architecture, where the encoders fuses a pretrained VGG network with an LSTM network that has a pretrained word embedding layer to encode the input. To generate the output, another LSTM network is used for decoding. When used with a pretrained VGG network, the VGG-Seq2Seq model managed to achieve reasonable results with 0.06, 0.12, 0.03 BLEU, WBSS and CBSS, respectively. Moreover, the VGG-Seq2Seq is not expensive to train.

KW - Global Vectors for Word Representation

KW - Sequence to sequence

KW - VGG Network

UR - https://www.scopus.com/pages/publications/85051091831

M3 - Conference article

AN - SCOPUS:85051091831

SN - 1613-0073

VL - 2125

JO - CEUR Workshop Proceedings

JF - CEUR Workshop Proceedings

Y2 - 10 September 2018 through 14 September 2018

ER -

JUST at VQA-Med: A VGG-Seq2Seq model

Abstract

Keywords

Other files and links

Fingerprint

Cite this