Micro-Video Popularity Prediction with Bidirectional Deep Encoding Network

Peiguang Jing; Xuqing Ye; Yu Liu; Yuting Su

doi:10.3788/LOP202259.0811009

Journals >Laser & Optoelectronics Progress >Volume 59 >Issue 8 >Page 0811009 > Article

Laser & Optoelectronics Progress
Vol. 59, Issue 8, 0811009 (2022)

Peiguang Jing¹, Xuqing Ye^1、*, Yu Liu², and Yuting Su¹

Author Affiliations

¹School of Electrical and Information Engineering, Tianjin University, Tianjin 300072, China

²School of Microelectronics, Tianjin University, Tianjin 300072, China

show less

DOI: 10.3788/LOP202259.0811009 Cite this Article Set citation alerts

Peiguang Jing, Xuqing Ye, Yu Liu, Yuting Su. Micro-Video Popularity Prediction with Bidirectional Deep Encoding Network[J]. Laser & Optoelectronics Progress, 2022, 59(8): 0811009 Copy Citation Text

show less

Fig. 1. Illustration of proposed model

Download full size

Fig. 2. nMSE changes with model iterations

Download full size

Fig. 3. Influence of

λ

on performance of proposed model

Download full size

Parameter	Top50	Top100	Top200	Bottom200	Bottom100	Bottom50
NP	0.312	0.261	0.222	0.152	0.151	0.150
NF	0.344	0.260	0.224	0.151	0.150	0.150
TNV	0.256	0.272	0.222	0.151	0.150	0.150
TPS	0.402	0.295	0.226	0.151	0.150	0.150

Table 1. Impact of social cues on popularity scores

Experiment	Experiment 1	Experiment 2	Experiment 3	Experiment 4	Experiment 5
nMSE	0.984	0.956	0.952	0.944	0.921
Self-attention	×	√	√	×	√
USM	×	×	√	√	√
ULGM	×	×	×	√	√

Table 2. Performance comparison of modules involved in framework

Visual feature	Color	Object	SentiBank	Aesthetic	All
nMSE	0.951	0.946	0.944	0.947	0.921
Top50	0.326	0.341	0.346	0.332	0.352
Top100	0.231	0.267	0.262	0.265	0.292
Top200	0.228	0.249	0.246	0.241	0.266
Bottom200	0.209	0.192	0.195	0.201	0.213
Bottom100	0.196	0.190	0.190	0.197	0.208
Bottom50	0.190	0.185	0.183	0.181	0.204

Table 3. Performance comparison among different visual features

Visual feature	V+A+T	V+A+S	V+T+S	A+T+S	V+A+T+S
nMSE	0.951	0.930	0.936	0.948	0.921
Top50	0.389	0.340	0.352	0.365	0.352
Top100	0.273	0.288	0.297	0.271	0.292
Top200	0.223	0.261	0.264	0.245	0.266
Bottom200	0.200	0.196	0.194	0.194	0.213
Bottom100	0.205	0.192	0.189	0.180	0.208
Bottom50	0.204	0.187	0.177	0.172	0.204

Table 4. Performance comparison among different modal characteristics

Method	nMSE
MLHR	1.167
MSNL	1.098
MvDA	0.982
ELM	0.982
TMALL	0.979
TLRMVR	0.934
Ours	0.921

Table 5. Performance comparison between our proposed method and several methods

Peiguang Jing, Xuqing Ye, Yu Liu, Yuting Su. Micro-Video Popularity Prediction with Bidirectional Deep Encoding Network[J]. Laser & Optoelectronics Progress, 2022, 59(8): 0811009

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information