Abstract
| Original language | English |
|---|---|
| Title of host publication | Image and Video Technology, PSIVT 2015 |
| Editors | Thomas Bräunl, Brendan McCane, Mariano Rivera, Xinguo Yu |
| Place of Publication | USA |
| Publisher | Springer-Verlag London Ltd. |
| Pages | 631-641 |
| Number of pages | 11 |
| Volume | 9431 |
| ISBN (Electronic) | 9783319294513 |
| ISBN (Print) | 9783319294506 |
| DOIs | |
| Publication status | Published - 2016 |
| Event | 7th Pacific-Rim Symposium on Image and Video Technology: PSIVT 2015 - Auckland, New Zealand Duration: 23 Nov 2015 → 27 Nov 2015 |
Publication series
| Name | Lecture Notes in Computer Science |
|---|
Conference
| Conference | 7th Pacific-Rim Symposium on Image and Video Technology |
|---|---|
| Abbreviated title | PSIVT 2015 |
| Country | New Zealand |
| City | Auckland |
| Period | 23/11/15 → 27/11/15 |
Fingerprint
Cite this
}
Deep Boltzmann Machines for i-Vector Based Audio-Visual Person Identification. / Alam, M.R. ; Bennamoun, Mohammed; Togneri, Roberto; Sohel, Ferdous.
Image and Video Technology, PSIVT 2015. ed. / Thomas Bräunl; Brendan McCane; Mariano Rivera; Xinguo Yu. Vol. 9431 USA : Springer-Verlag London Ltd., 2016. p. 631-641 (Lecture Notes in Computer Science).Research output: Chapter in Book/Conference paper › Conference paper
TY - GEN
T1 - Deep Boltzmann Machines for i-Vector Based Audio-Visual Person Identification
AU - Alam, M.R.
AU - Bennamoun, Mohammed
AU - Togneri, Roberto
AU - Sohel, Ferdous
PY - 2016
Y1 - 2016
N2 - We propose an approach using DBM-DNNs for i-vector based audio-visual person identification. The unsupervised training of two Deep Boltzmann Machines DBMspeech and DBMface is performed using unlabeled audio and visual data from a set of background subjects. The DBMs are then used to initialize two corresponding DNNs for classification, referred to as the DBM-DNNspeech and DBM-DNNface in this paper. The DBM-DNNs are discriminatively fine-tuned using the back- propagation on a set of training data and evaluated on a set of test data from the target subjects. We compared their performance with the cosine distance (cosDist) and the state-of-the-art DBN-DNN classifier. We also tested three different configurations of the DBM-DNNs. We show that DBM-DNNs with two hidden layers and 800 units in each hidden layer achieved best identification performance for 400 dimensional i-vectors as input. Our experiments were carried out on the challenging MOBIO dataset. © Springer International Publishing Switzerland 2016.
AB - We propose an approach using DBM-DNNs for i-vector based audio-visual person identification. The unsupervised training of two Deep Boltzmann Machines DBMspeech and DBMface is performed using unlabeled audio and visual data from a set of background subjects. The DBMs are then used to initialize two corresponding DNNs for classification, referred to as the DBM-DNNspeech and DBM-DNNface in this paper. The DBM-DNNs are discriminatively fine-tuned using the back- propagation on a set of training data and evaluated on a set of test data from the target subjects. We compared their performance with the cosine distance (cosDist) and the state-of-the-art DBN-DNN classifier. We also tested three different configurations of the DBM-DNNs. We show that DBM-DNNs with two hidden layers and 800 units in each hidden layer achieved best identification performance for 400 dimensional i-vectors as input. Our experiments were carried out on the challenging MOBIO dataset. © Springer International Publishing Switzerland 2016.
U2 - 10.1007/978-3-319-29451-3_50
DO - 10.1007/978-3-319-29451-3_50
M3 - Conference paper
SN - 9783319294506
VL - 9431
T3 - Lecture Notes in Computer Science
SP - 631
EP - 641
BT - Image and Video Technology, PSIVT 2015
A2 - Bräunl, Thomas
A2 - McCane, Brendan
A2 - Rivera, Mariano
A2 - Yu, Xinguo
PB - Springer-Verlag London Ltd.
CY - USA
ER -