Page 160 - My FlipBook
P. 160
研
人 Research Fellow
員 呂俊賢 Chun-Shien Lu
Faculty Ph.D., Electrical Engineering, National Cheng-Kung University, Taiwan
T +886-2-2788-3799 ext. 1513 E lcs@iis.sinica.edu.tw
F +886-2-2782-4814 W www.iis.sinica.edu.tw/pages/lcs
・ Research Fellow, Institute of Information Science, Academia Sinica (2013/3-present)
・ Deputy Director, Research Center for Information Technology Innovation, Academia
Sinica (2015/4-2020/1)
・ Associate editor of IEEE Trans. on Image Processing (2010/12-2014) (2018/3-present)
・ Ta-You Wu Memorial Award, Ministry of Science and Technology, Taiwan, ROC (2007)
Research Description
My recent research interests are in the (Deep) Compressive Sensing and AI Security & Privacy.
For compressive sensing (CS), it is a new paradigm of simultaneous sampling and compression. Without being restricted to the constraint
of Nyquist rate, CS can, in theory, perfectly reconstruct the original signal under the constraints that only a few samples or measurements
extracted from an original signal are needed and the signal is sparse. The unique characteristic of CS is that sampling and compression can
be simultaneously achieved for use in resource-limited mobile devices. Nevertheless, there are several aspects needed to be considered in
CS, including the design of sensing matrix and dictionary/basis, the sparsity or low-rank of signals, and time-consuming optimization for
signal recovery. Thus, our motivation is to study the e cient and exible design of compressive sensing based on deep learning, called "Deep
Sensing."
For arti cial intelligence (AI) security and privacy, due to the population and development of deep learning technologies, recent researches
reveal that the sophisticated design of adversarial examples (inputs) can achieve efficient fooling effect on well-trained deep neural
networks (DNNs). Meanwhile, the "adversarial perturbations" introduced by adversarial samples are indistinguishable from the benign
inputs in terms of human perception. According to the literature, the amount of publications, pertaining to AI Security and Privacy, has
been grown exponentially since 2014. This indicates that the issues of AI security and privacy has received much attention recently. In view
of the observations regarding the relationship among AI Model, Security, Privacy, and Data Hiding, we study the issues of balancing model
accuracy, AI security, and AI privacy.
Publications
1. Yi-Lin Sung, Sung-Hsien Hsieh, Soo-Chang Pei, and Chun- 6. Chia-Mu Yu, Chun-Shien Lu, and Sy-Yen Kuo, "Compressed
Shien Lu, "Difference-Seeking Generative Adversarial Network- Sensing-Based Clone Identification in Sensor Network," IEEE
-Unseen Sample Generation," International Conference on Trans. on Wireless Communications , Vol. 15, No. 4, pp. 3071-
Learning Representations (ICLR), Addis Ababa, Ethiopia, April 3084, 2016.
26-30, 2020. (full paper)
7. Yao-Tung Tsou, Chun-Shien Lu, and Sy-Yen Kuo, "MoteSec-
2. Chia-Mu, Yu, Sarada Gochhayat, Mauro Conti, and Chun-Shien, Aware: A Practical Secure Mechanism for Wireless Sensor
Lu, "Privacy Aware Data Deduplication for Side Channel in Networks," IEEE Trans. on Wireless Communications, Vol. 12,
Cloud Storage," Accepted and to Appear in IEEE Trans. on Cloud No. 6, pp. 2817-2829, 2013.
Computing .
8. Chia-Mu Yu, Yao-Tung Tsou, Chun-Shien Lu, and Sy-Yen
3. Sung-Hsien Hsieh, Wei-Jie, Liang, Chun-Shien Lu, and Soo- Kuo, "Localized Algorithms for Detection of Node Replication
Chang Pei, "Distributed Compressive Sensing: Performance Attacks in Mobile Sensor Networks," IEEE Trans. on Information
Analysis with Diverse Signal Ensembles,'' Accepted and to Forensics, and Security, Vol. 8, No. 5, pp. 754-768, 2013.
Appear in IEEE Trans. on Signal Processing.
9. Chao-Yung Hsu, Chun-Shien Lu, and Soo-Chang Pei, "Image
4. Sung-Hsien Hsieh, Chun-Shien Lu, and Soo-Chang Pei, Feature Extraction in Encrypted Domain with Privacy-Preserving
"Compressive Sensing Matrix Design for Fast Encoding and SIFT," IEEE Trans. on Image Processing, Vol. 21, No. 11, pp.
Decoding via Sparse Fourier Transform," IEEE Signal Processing 4593-4607, 2012.
Letters, Vol. 25, No. 4, pp. 591-595, 2018.
10. Chun-Shien Lu and Chao-Yung Hsu, "Constraint-Optimized
5. Wei-Jie, Liang, Gang-Xuan Lin, and Chun-Shien Lu, "Tree Keypoint Removal/Insertion Attack: Security Threat to Scale-
Structure Sparsity Pattern Guided Convex Optimization for Space Image Feature Extraction," ACM Multimedia Conference
Compressive Sensing of Large-Scale Images," IEEE Trans. on (ACM MM), Oct. 30-Nov. 02, Nara, Japan, pp. 629-638, 2012.
Image Processing, Vol. 26, No. 2, pp. 847-859, 2017. (full paper, acceptance rate 20.2%).
158
人 Research Fellow
員 呂俊賢 Chun-Shien Lu
Faculty Ph.D., Electrical Engineering, National Cheng-Kung University, Taiwan
T +886-2-2788-3799 ext. 1513 E lcs@iis.sinica.edu.tw
F +886-2-2782-4814 W www.iis.sinica.edu.tw/pages/lcs
・ Research Fellow, Institute of Information Science, Academia Sinica (2013/3-present)
・ Deputy Director, Research Center for Information Technology Innovation, Academia
Sinica (2015/4-2020/1)
・ Associate editor of IEEE Trans. on Image Processing (2010/12-2014) (2018/3-present)
・ Ta-You Wu Memorial Award, Ministry of Science and Technology, Taiwan, ROC (2007)
Research Description
My recent research interests are in the (Deep) Compressive Sensing and AI Security & Privacy.
For compressive sensing (CS), it is a new paradigm of simultaneous sampling and compression. Without being restricted to the constraint
of Nyquist rate, CS can, in theory, perfectly reconstruct the original signal under the constraints that only a few samples or measurements
extracted from an original signal are needed and the signal is sparse. The unique characteristic of CS is that sampling and compression can
be simultaneously achieved for use in resource-limited mobile devices. Nevertheless, there are several aspects needed to be considered in
CS, including the design of sensing matrix and dictionary/basis, the sparsity or low-rank of signals, and time-consuming optimization for
signal recovery. Thus, our motivation is to study the e cient and exible design of compressive sensing based on deep learning, called "Deep
Sensing."
For arti cial intelligence (AI) security and privacy, due to the population and development of deep learning technologies, recent researches
reveal that the sophisticated design of adversarial examples (inputs) can achieve efficient fooling effect on well-trained deep neural
networks (DNNs). Meanwhile, the "adversarial perturbations" introduced by adversarial samples are indistinguishable from the benign
inputs in terms of human perception. According to the literature, the amount of publications, pertaining to AI Security and Privacy, has
been grown exponentially since 2014. This indicates that the issues of AI security and privacy has received much attention recently. In view
of the observations regarding the relationship among AI Model, Security, Privacy, and Data Hiding, we study the issues of balancing model
accuracy, AI security, and AI privacy.
Publications
1. Yi-Lin Sung, Sung-Hsien Hsieh, Soo-Chang Pei, and Chun- 6. Chia-Mu Yu, Chun-Shien Lu, and Sy-Yen Kuo, "Compressed
Shien Lu, "Difference-Seeking Generative Adversarial Network- Sensing-Based Clone Identification in Sensor Network," IEEE
-Unseen Sample Generation," International Conference on Trans. on Wireless Communications , Vol. 15, No. 4, pp. 3071-
Learning Representations (ICLR), Addis Ababa, Ethiopia, April 3084, 2016.
26-30, 2020. (full paper)
7. Yao-Tung Tsou, Chun-Shien Lu, and Sy-Yen Kuo, "MoteSec-
2. Chia-Mu, Yu, Sarada Gochhayat, Mauro Conti, and Chun-Shien, Aware: A Practical Secure Mechanism for Wireless Sensor
Lu, "Privacy Aware Data Deduplication for Side Channel in Networks," IEEE Trans. on Wireless Communications, Vol. 12,
Cloud Storage," Accepted and to Appear in IEEE Trans. on Cloud No. 6, pp. 2817-2829, 2013.
Computing .
8. Chia-Mu Yu, Yao-Tung Tsou, Chun-Shien Lu, and Sy-Yen
3. Sung-Hsien Hsieh, Wei-Jie, Liang, Chun-Shien Lu, and Soo- Kuo, "Localized Algorithms for Detection of Node Replication
Chang Pei, "Distributed Compressive Sensing: Performance Attacks in Mobile Sensor Networks," IEEE Trans. on Information
Analysis with Diverse Signal Ensembles,'' Accepted and to Forensics, and Security, Vol. 8, No. 5, pp. 754-768, 2013.
Appear in IEEE Trans. on Signal Processing.
9. Chao-Yung Hsu, Chun-Shien Lu, and Soo-Chang Pei, "Image
4. Sung-Hsien Hsieh, Chun-Shien Lu, and Soo-Chang Pei, Feature Extraction in Encrypted Domain with Privacy-Preserving
"Compressive Sensing Matrix Design for Fast Encoding and SIFT," IEEE Trans. on Image Processing, Vol. 21, No. 11, pp.
Decoding via Sparse Fourier Transform," IEEE Signal Processing 4593-4607, 2012.
Letters, Vol. 25, No. 4, pp. 591-595, 2018.
10. Chun-Shien Lu and Chao-Yung Hsu, "Constraint-Optimized
5. Wei-Jie, Liang, Gang-Xuan Lin, and Chun-Shien Lu, "Tree Keypoint Removal/Insertion Attack: Security Threat to Scale-
Structure Sparsity Pattern Guided Convex Optimization for Space Image Feature Extraction," ACM Multimedia Conference
Compressive Sensing of Large-Scale Images," IEEE Trans. on (ACM MM), Oct. 30-Nov. 02, Nara, Japan, pp. 629-638, 2012.
Image Processing, Vol. 26, No. 2, pp. 847-859, 2017. (full paper, acceptance rate 20.2%).
158
