Sorry, you need to enable JavaScript to visit this website.

Machine Learning for Signal Processing

Cross-Modality Distillation: A Case for Conditional Generative Adversarial Networks


In this paper, we propose to use a Conditional Generative Adversarial Network (CGAN) for distilling (i.e. transferring) knowledge from sensor data and enhancing low-resolution target detection. In unconstrained surveillance settings, sensor measurements are often noisy, degraded, corrupted, and even missing/absent, thereby presenting a significant problem for multi-modal fusion. We therefore specifically tackle the problem of a missing modality in our attempt to propose an algorithm

Paper Details

Authors:
Siddharth Roheda, Benjamin S Riggan, Hamid Krim, Liyi Dai
Submitted On:
13 April 2018 - 12:56pm
Short Link:
Type:
Event:
Paper Code:
Document Year:
Cite

Document Files

conference_poster_sroheda_4.pdf

(123)

Subscribe

[1] Siddharth Roheda, Benjamin S Riggan, Hamid Krim, Liyi Dai, "Cross-Modality Distillation: A Case for Conditional Generative Adversarial Networks", IEEE SigPort, 2018. [Online]. Available: http://sigport.org/2729. Accessed: May. 23, 2019.
@article{2729-18,
url = {http://sigport.org/2729},
author = {Siddharth Roheda; Benjamin S Riggan; Hamid Krim; Liyi Dai },
publisher = {IEEE SigPort},
title = {Cross-Modality Distillation: A Case for Conditional Generative Adversarial Networks},
year = {2018} }
TY - EJOUR
T1 - Cross-Modality Distillation: A Case for Conditional Generative Adversarial Networks
AU - Siddharth Roheda; Benjamin S Riggan; Hamid Krim; Liyi Dai
PY - 2018
PB - IEEE SigPort
UR - http://sigport.org/2729
ER -
Siddharth Roheda, Benjamin S Riggan, Hamid Krim, Liyi Dai. (2018). Cross-Modality Distillation: A Case for Conditional Generative Adversarial Networks. IEEE SigPort. http://sigport.org/2729
Siddharth Roheda, Benjamin S Riggan, Hamid Krim, Liyi Dai, 2018. Cross-Modality Distillation: A Case for Conditional Generative Adversarial Networks. Available at: http://sigport.org/2729.
Siddharth Roheda, Benjamin S Riggan, Hamid Krim, Liyi Dai. (2018). "Cross-Modality Distillation: A Case for Conditional Generative Adversarial Networks." Web.
1. Siddharth Roheda, Benjamin S Riggan, Hamid Krim, Liyi Dai. Cross-Modality Distillation: A Case for Conditional Generative Adversarial Networks [Internet]. IEEE SigPort; 2018. Available from : http://sigport.org/2729

DIRECTLY SOLVING THE ORIGINAL RATIOCUT PROBLEM FOR EFFECTIVE DATA CLUSTERING

Paper Details

Authors:
Jing Li, Feiping Nie, Xuelong Li
Submitted On:
13 April 2018 - 10:58am
Short Link:
Type:
Event:

Document Files

ICASSP 2018.pdf

(14)

Subscribe

[1] Jing Li, Feiping Nie, Xuelong Li, "DIRECTLY SOLVING THE ORIGINAL RATIOCUT PROBLEM FOR EFFECTIVE DATA CLUSTERING", IEEE SigPort, 2018. [Online]. Available: http://sigport.org/2714. Accessed: May. 23, 2019.
@article{2714-18,
url = {http://sigport.org/2714},
author = {Jing Li; Feiping Nie; Xuelong Li },
publisher = {IEEE SigPort},
title = {DIRECTLY SOLVING THE ORIGINAL RATIOCUT PROBLEM FOR EFFECTIVE DATA CLUSTERING},
year = {2018} }
TY - EJOUR
T1 - DIRECTLY SOLVING THE ORIGINAL RATIOCUT PROBLEM FOR EFFECTIVE DATA CLUSTERING
AU - Jing Li; Feiping Nie; Xuelong Li
PY - 2018
PB - IEEE SigPort
UR - http://sigport.org/2714
ER -
Jing Li, Feiping Nie, Xuelong Li. (2018). DIRECTLY SOLVING THE ORIGINAL RATIOCUT PROBLEM FOR EFFECTIVE DATA CLUSTERING. IEEE SigPort. http://sigport.org/2714
Jing Li, Feiping Nie, Xuelong Li, 2018. DIRECTLY SOLVING THE ORIGINAL RATIOCUT PROBLEM FOR EFFECTIVE DATA CLUSTERING. Available at: http://sigport.org/2714.
Jing Li, Feiping Nie, Xuelong Li. (2018). "DIRECTLY SOLVING THE ORIGINAL RATIOCUT PROBLEM FOR EFFECTIVE DATA CLUSTERING." Web.
1. Jing Li, Feiping Nie, Xuelong Li. DIRECTLY SOLVING THE ORIGINAL RATIOCUT PROBLEM FOR EFFECTIVE DATA CLUSTERING [Internet]. IEEE SigPort; 2018. Available from : http://sigport.org/2714

Regressing Kernel Dictionary Learning


In this paper, we present a kernelized dictionary learning framework for carrying out regression to model signals having a complex non-linear nature. A joint optimization is carried out where the regression weights are learnt together with the dictionary and coefficients. Relevant formulation and dictionary building steps are provided. To demonstrate the effectiveness of the proposed technique, elaborate experimental results using different real-life datasets are presented.

Paper Details

Authors:
Submitted On:
25 April 2018 - 5:20am
Short Link:
Type:
Event:
Presenter's Name:
Paper Code:
Document Year:
Cite

Document Files

Poster_v31.pdf

(111)

Subscribe

[1] , "Regressing Kernel Dictionary Learning", IEEE SigPort, 2018. [Online]. Available: http://sigport.org/2711. Accessed: May. 23, 2019.
@article{2711-18,
url = {http://sigport.org/2711},
author = { },
publisher = {IEEE SigPort},
title = {Regressing Kernel Dictionary Learning},
year = {2018} }
TY - EJOUR
T1 - Regressing Kernel Dictionary Learning
AU -
PY - 2018
PB - IEEE SigPort
UR - http://sigport.org/2711
ER -
. (2018). Regressing Kernel Dictionary Learning. IEEE SigPort. http://sigport.org/2711
, 2018. Regressing Kernel Dictionary Learning. Available at: http://sigport.org/2711.
. (2018). "Regressing Kernel Dictionary Learning." Web.
1. . Regressing Kernel Dictionary Learning [Internet]. IEEE SigPort; 2018. Available from : http://sigport.org/2711

LOCALITY-PRESERVING COMPLEX-VALUED GAUSSIAN PROCESS LATENT VARIABLE MODEL FOR ROBUST FACE RECOGNITION

Paper Details

Authors:
Submitted On:
13 April 2018 - 2:03am
Short Link:
Type:
Event:

Document Files

ICASSP2018face_poster.pdf

(94)

Subscribe

[1] , "LOCALITY-PRESERVING COMPLEX-VALUED GAUSSIAN PROCESS LATENT VARIABLE MODEL FOR ROBUST FACE RECOGNITION", IEEE SigPort, 2018. [Online]. Available: http://sigport.org/2617. Accessed: May. 23, 2019.
@article{2617-18,
url = {http://sigport.org/2617},
author = { },
publisher = {IEEE SigPort},
title = {LOCALITY-PRESERVING COMPLEX-VALUED GAUSSIAN PROCESS LATENT VARIABLE MODEL FOR ROBUST FACE RECOGNITION},
year = {2018} }
TY - EJOUR
T1 - LOCALITY-PRESERVING COMPLEX-VALUED GAUSSIAN PROCESS LATENT VARIABLE MODEL FOR ROBUST FACE RECOGNITION
AU -
PY - 2018
PB - IEEE SigPort
UR - http://sigport.org/2617
ER -
. (2018). LOCALITY-PRESERVING COMPLEX-VALUED GAUSSIAN PROCESS LATENT VARIABLE MODEL FOR ROBUST FACE RECOGNITION. IEEE SigPort. http://sigport.org/2617
, 2018. LOCALITY-PRESERVING COMPLEX-VALUED GAUSSIAN PROCESS LATENT VARIABLE MODEL FOR ROBUST FACE RECOGNITION. Available at: http://sigport.org/2617.
. (2018). "LOCALITY-PRESERVING COMPLEX-VALUED GAUSSIAN PROCESS LATENT VARIABLE MODEL FOR ROBUST FACE RECOGNITION." Web.
1. . LOCALITY-PRESERVING COMPLEX-VALUED GAUSSIAN PROCESS LATENT VARIABLE MODEL FOR ROBUST FACE RECOGNITION [Internet]. IEEE SigPort; 2018. Available from : http://sigport.org/2617

Discriminative Probabilistic Framework for Generalized Multi-Instance Learning


Multiple-instance learning is a framework for learning from data consisting of bags of instances labeled at the bag level. A common assumption in multi-instance learning is that a bag label is positive if and only if at least one instance in the bag is positive. In practice, this assumption may be violated. For example, experts may provide a noisy label to a bag consisting of many instances, to reduce labeling time.

Paper Details

Authors:
Anh T Pham, Raviv Raich, Xiaoli Fern; Weng K Wong; Xinze Guan
Submitted On:
25 April 2018 - 1:39pm
Short Link:
Type:
Event:
Presenter's Name:
Paper Code:
Document Year:
Cite

Document Files

Discriminative Probabilistic Framework for Generalized Multi-Instance_ICASSP2018.pdf

(10)

Subscribe

[1] Anh T Pham, Raviv Raich, Xiaoli Fern; Weng K Wong; Xinze Guan, "Discriminative Probabilistic Framework for Generalized Multi-Instance Learning", IEEE SigPort, 2018. [Online]. Available: http://sigport.org/2512. Accessed: May. 23, 2019.
@article{2512-18,
url = {http://sigport.org/2512},
author = {Anh T Pham; Raviv Raich; Xiaoli Fern; Weng K Wong; Xinze Guan },
publisher = {IEEE SigPort},
title = {Discriminative Probabilistic Framework for Generalized Multi-Instance Learning},
year = {2018} }
TY - EJOUR
T1 - Discriminative Probabilistic Framework for Generalized Multi-Instance Learning
AU - Anh T Pham; Raviv Raich; Xiaoli Fern; Weng K Wong; Xinze Guan
PY - 2018
PB - IEEE SigPort
UR - http://sigport.org/2512
ER -
Anh T Pham, Raviv Raich, Xiaoli Fern; Weng K Wong; Xinze Guan. (2018). Discriminative Probabilistic Framework for Generalized Multi-Instance Learning. IEEE SigPort. http://sigport.org/2512
Anh T Pham, Raviv Raich, Xiaoli Fern; Weng K Wong; Xinze Guan, 2018. Discriminative Probabilistic Framework for Generalized Multi-Instance Learning. Available at: http://sigport.org/2512.
Anh T Pham, Raviv Raich, Xiaoli Fern; Weng K Wong; Xinze Guan. (2018). "Discriminative Probabilistic Framework for Generalized Multi-Instance Learning." Web.
1. Anh T Pham, Raviv Raich, Xiaoli Fern; Weng K Wong; Xinze Guan. Discriminative Probabilistic Framework for Generalized Multi-Instance Learning [Internet]. IEEE SigPort; 2018. Available from : http://sigport.org/2512

Speaker-Invariant Training via Adversarial Learning


We propose a novel adversarial multi-task learning scheme, aiming at actively curtailing the inter-talker feature variability while maximizing its senone discriminability so as to enhance the performance of a deep neural network (DNN) based ASR system. We call the scheme speaker-invariant training (SIT). In SIT, a DNN acoustic model and a speaker classifier network are jointly optimized to minimize the senone (tied triphone state) classification loss, and simultaneously mini-maximize the speaker classification loss.

Paper Details

Authors:
Zhong Meng, Jinyu Li, Zhuo Chen, Yong Zhao, Vadim Mazalov, Yifan Gong, Biing-Hwang (Fred) Juang
Submitted On:
12 May 2019 - 9:29pm
Short Link:
Type:
Event:
Presenter's Name:
Paper Code:
Document Year:
Cite

Document Files

sit_poster.pptx

(137)

Subscribe

[1] Zhong Meng, Jinyu Li, Zhuo Chen, Yong Zhao, Vadim Mazalov, Yifan Gong, Biing-Hwang (Fred) Juang, "Speaker-Invariant Training via Adversarial Learning", IEEE SigPort, 2018. [Online]. Available: http://sigport.org/2508. Accessed: May. 23, 2019.
@article{2508-18,
url = {http://sigport.org/2508},
author = {Zhong Meng; Jinyu Li; Zhuo Chen; Yong Zhao; Vadim Mazalov; Yifan Gong; Biing-Hwang (Fred) Juang },
publisher = {IEEE SigPort},
title = {Speaker-Invariant Training via Adversarial Learning},
year = {2018} }
TY - EJOUR
T1 - Speaker-Invariant Training via Adversarial Learning
AU - Zhong Meng; Jinyu Li; Zhuo Chen; Yong Zhao; Vadim Mazalov; Yifan Gong; Biing-Hwang (Fred) Juang
PY - 2018
PB - IEEE SigPort
UR - http://sigport.org/2508
ER -
Zhong Meng, Jinyu Li, Zhuo Chen, Yong Zhao, Vadim Mazalov, Yifan Gong, Biing-Hwang (Fred) Juang. (2018). Speaker-Invariant Training via Adversarial Learning. IEEE SigPort. http://sigport.org/2508
Zhong Meng, Jinyu Li, Zhuo Chen, Yong Zhao, Vadim Mazalov, Yifan Gong, Biing-Hwang (Fred) Juang, 2018. Speaker-Invariant Training via Adversarial Learning. Available at: http://sigport.org/2508.
Zhong Meng, Jinyu Li, Zhuo Chen, Yong Zhao, Vadim Mazalov, Yifan Gong, Biing-Hwang (Fred) Juang. (2018). "Speaker-Invariant Training via Adversarial Learning." Web.
1. Zhong Meng, Jinyu Li, Zhuo Chen, Yong Zhao, Vadim Mazalov, Yifan Gong, Biing-Hwang (Fred) Juang. Speaker-Invariant Training via Adversarial Learning [Internet]. IEEE SigPort; 2018. Available from : http://sigport.org/2508

Adversarial Teacher-Student Learning for Unsupervised Adaptation


The teacher-student (T/S) learning has been shown effective in unsupervised domain adaptation ts_adapt. It is a form of transfer learning, not in terms of the transfer of recognition decisions, but the knowledge of posteriori probabilities in the source domain as evaluated by the teacher model. It learns to handle the speaker and environment variability inherent in and restricted to the speech signal in the target domain without proactively addressing the robustness to other likely conditions. Performance degradation may thus ensue.

Paper Details

Authors:
Zhong Meng, Jinyu Li, Yifan Gong, Biing-Hwang (Fred) Juang
Submitted On:
12 May 2019 - 9:31pm
Short Link:
Type:
Event:
Presenter's Name:
Paper Code:
Document Year:
Cite

Document Files

ats_poster_v2.pptx

(86)

Subscribe

[1] Zhong Meng, Jinyu Li, Yifan Gong, Biing-Hwang (Fred) Juang, "Adversarial Teacher-Student Learning for Unsupervised Adaptation", IEEE SigPort, 2018. [Online]. Available: http://sigport.org/2506. Accessed: May. 23, 2019.
@article{2506-18,
url = {http://sigport.org/2506},
author = {Zhong Meng; Jinyu Li; Yifan Gong; Biing-Hwang (Fred) Juang },
publisher = {IEEE SigPort},
title = {Adversarial Teacher-Student Learning for Unsupervised Adaptation},
year = {2018} }
TY - EJOUR
T1 - Adversarial Teacher-Student Learning for Unsupervised Adaptation
AU - Zhong Meng; Jinyu Li; Yifan Gong; Biing-Hwang (Fred) Juang
PY - 2018
PB - IEEE SigPort
UR - http://sigport.org/2506
ER -
Zhong Meng, Jinyu Li, Yifan Gong, Biing-Hwang (Fred) Juang. (2018). Adversarial Teacher-Student Learning for Unsupervised Adaptation. IEEE SigPort. http://sigport.org/2506
Zhong Meng, Jinyu Li, Yifan Gong, Biing-Hwang (Fred) Juang, 2018. Adversarial Teacher-Student Learning for Unsupervised Adaptation. Available at: http://sigport.org/2506.
Zhong Meng, Jinyu Li, Yifan Gong, Biing-Hwang (Fred) Juang. (2018). "Adversarial Teacher-Student Learning for Unsupervised Adaptation." Web.
1. Zhong Meng, Jinyu Li, Yifan Gong, Biing-Hwang (Fred) Juang. Adversarial Teacher-Student Learning for Unsupervised Adaptation [Internet]. IEEE SigPort; 2018. Available from : http://sigport.org/2506

A Fully Convolutional Tri-branch Network (FCTN) For Domain Adaptation


A domain adaptation method for urban scene segmentation is proposed in this work. We develop a fully convolutional tri-branch network, where two branches assign pseudo labels to images in the unlabeled target domain while the third branch is trained with supervision based on images in the pseudo-labeled target domain. The re-labeling and re-training processes alternate. With this design, the tri-branch network learns target-specific discriminative representations progressively and, as a result, the cross-domain capability of the segmenter improves.

Paper Details

Authors:
Junting Zhang, Chen Liang, C.-C. Jay Kuo
Submitted On:
12 April 2018 - 5:32pm
Short Link:
Type:
Event:
Presenter's Name:
Paper Code:
Document Year:
Cite

Document Files

A Fully Convolutional Tri-branch Network (FCTN) For Domain Adaptation-Poster

(111)

Keywords

Additional Categories

Subscribe

[1] Junting Zhang, Chen Liang, C.-C. Jay Kuo, "A Fully Convolutional Tri-branch Network (FCTN) For Domain Adaptation", IEEE SigPort, 2018. [Online]. Available: http://sigport.org/2503. Accessed: May. 23, 2019.
@article{2503-18,
url = {http://sigport.org/2503},
author = {Junting Zhang; Chen Liang; C.-C. Jay Kuo },
publisher = {IEEE SigPort},
title = {A Fully Convolutional Tri-branch Network (FCTN) For Domain Adaptation},
year = {2018} }
TY - EJOUR
T1 - A Fully Convolutional Tri-branch Network (FCTN) For Domain Adaptation
AU - Junting Zhang; Chen Liang; C.-C. Jay Kuo
PY - 2018
PB - IEEE SigPort
UR - http://sigport.org/2503
ER -
Junting Zhang, Chen Liang, C.-C. Jay Kuo. (2018). A Fully Convolutional Tri-branch Network (FCTN) For Domain Adaptation. IEEE SigPort. http://sigport.org/2503
Junting Zhang, Chen Liang, C.-C. Jay Kuo, 2018. A Fully Convolutional Tri-branch Network (FCTN) For Domain Adaptation. Available at: http://sigport.org/2503.
Junting Zhang, Chen Liang, C.-C. Jay Kuo. (2018). "A Fully Convolutional Tri-branch Network (FCTN) For Domain Adaptation." Web.
1. Junting Zhang, Chen Liang, C.-C. Jay Kuo. A Fully Convolutional Tri-branch Network (FCTN) For Domain Adaptation [Internet]. IEEE SigPort; 2018. Available from : http://sigport.org/2503

Pattern Localization in Time Series through Signal-To-Model Alignment in Latent Space


In this paper, we study the problem of locating a predefined sequence of patterns in a time series. In particular, the studied scenario assumes a theoretical model is available that contains the expected locations of the patterns. This problem is found in several contexts, and it is commonly solved by first synthesizing a time series from the model, and then aligning it to the true time series through dynamic time warping. We propose a technique that increases the similarity of both time series before aligning them, by mapping them into a latent correlation space.

Paper Details

Authors:
Steven Van Vaerenbergh, Ignacio Santamaría, Víctor Elvira, Matteo Salvatori
Submitted On:
12 April 2018 - 11:50am
Short Link:
Type:
Event:
Presenter's Name:
Paper Code:
Document Year:
Cite

Document Files

Pattern Localization in Time Series through Signal-To-Model Alignment in Latent Space

(370)

Subscribe

[1] Steven Van Vaerenbergh, Ignacio Santamaría, Víctor Elvira, Matteo Salvatori, "Pattern Localization in Time Series through Signal-To-Model Alignment in Latent Space", IEEE SigPort, 2018. [Online]. Available: http://sigport.org/2412. Accessed: May. 23, 2019.
@article{2412-18,
url = {http://sigport.org/2412},
author = {Steven Van Vaerenbergh; Ignacio Santamaría; Víctor Elvira; Matteo Salvatori },
publisher = {IEEE SigPort},
title = {Pattern Localization in Time Series through Signal-To-Model Alignment in Latent Space},
year = {2018} }
TY - EJOUR
T1 - Pattern Localization in Time Series through Signal-To-Model Alignment in Latent Space
AU - Steven Van Vaerenbergh; Ignacio Santamaría; Víctor Elvira; Matteo Salvatori
PY - 2018
PB - IEEE SigPort
UR - http://sigport.org/2412
ER -
Steven Van Vaerenbergh, Ignacio Santamaría, Víctor Elvira, Matteo Salvatori. (2018). Pattern Localization in Time Series through Signal-To-Model Alignment in Latent Space. IEEE SigPort. http://sigport.org/2412
Steven Van Vaerenbergh, Ignacio Santamaría, Víctor Elvira, Matteo Salvatori, 2018. Pattern Localization in Time Series through Signal-To-Model Alignment in Latent Space. Available at: http://sigport.org/2412.
Steven Van Vaerenbergh, Ignacio Santamaría, Víctor Elvira, Matteo Salvatori. (2018). "Pattern Localization in Time Series through Signal-To-Model Alignment in Latent Space." Web.
1. Steven Van Vaerenbergh, Ignacio Santamaría, Víctor Elvira, Matteo Salvatori. Pattern Localization in Time Series through Signal-To-Model Alignment in Latent Space [Internet]. IEEE SigPort; 2018. Available from : http://sigport.org/2412

Outlier-Robust Matrix Completion via lp-Minimization


Matrix completion refers to the recovery of a low‐rank matrix from only a subset of its possibly noisy entries, and has a variety of important applications such as collaborative filtering, image inpainting and restoration, system identification, node localization and genotype imputation. It is because many real-world signals can be approximated by a matrix whose rank is much smaller than the row and column numbers. Most techniques for matrix completion in the literature assume Gaussian noise and thus they are not robust to outliers.

Paper Details

Authors:
Wen-Jun Zeng, Hing Cheung So
Submitted On:
2 March 2018 - 1:57am
Short Link:
Type:

Document Files

rmp.pdf

(303)

Subscribe

[1] Wen-Jun Zeng, Hing Cheung So, "Outlier-Robust Matrix Completion via lp-Minimization", IEEE SigPort, 2018. [Online]. Available: http://sigport.org/2373. Accessed: May. 23, 2019.
@article{2373-18,
url = {http://sigport.org/2373},
author = {Wen-Jun Zeng; Hing Cheung So },
publisher = {IEEE SigPort},
title = {Outlier-Robust Matrix Completion via lp-Minimization},
year = {2018} }
TY - EJOUR
T1 - Outlier-Robust Matrix Completion via lp-Minimization
AU - Wen-Jun Zeng; Hing Cheung So
PY - 2018
PB - IEEE SigPort
UR - http://sigport.org/2373
ER -
Wen-Jun Zeng, Hing Cheung So. (2018). Outlier-Robust Matrix Completion via lp-Minimization. IEEE SigPort. http://sigport.org/2373
Wen-Jun Zeng, Hing Cheung So, 2018. Outlier-Robust Matrix Completion via lp-Minimization. Available at: http://sigport.org/2373.
Wen-Jun Zeng, Hing Cheung So. (2018). "Outlier-Robust Matrix Completion via lp-Minimization." Web.
1. Wen-Jun Zeng, Hing Cheung So. Outlier-Robust Matrix Completion via lp-Minimization [Internet]. IEEE SigPort; 2018. Available from : http://sigport.org/2373

Pages