Selected Papers:
Learning Sparse
Neural Networks via L0 and TL1 by a Relaxed Variable Splitting Method
with Application to Multi-scale Curve Classification (with F. Xue),
arXiv preprint arXiv: 1902.07419, Feb 20, 2019; to appear in Proc. World Congress Global Optimization, Metz, France, July, 2019.
A Study on Graph-Structured Recurrent Neural Networks and Sparsification with Application to Epidemic Forecasting (with Z. Li, X. Luo, B. Wang, A. Bertozzi ), arXiv preprint arXiv: 1902.05113, Feb. 14, 2019; to appear in Proc. World Congress Global Optimization, Metz, France, July 2019.
Convergence of
a Relaxed Variable Splitting Coarse Gradient Descent Method for Learning
Sparse Weight Binarized Activation Neural Networks (with T. Dinh),
arXiv preprint arXiv: 1901.09731, Jan 25, 2019.
AutoShuffleNet: Learning
Permutation Matrices via an Exact Lipschitz Continuous Penalty in
Deep Convolutional Neural Networks (with J. Lyu, S. Zhang, Y-Y. Qi),
arXiv preprint, arXiv: 1901.08624, Jan 24, 2019.
Convergence of a Relaxed
Variable Splitting Method for Learning Sparse Neural Networks via
L1, L0, and transformed-L1 Penalties (with T. Dinh), arXiv preprint,
arXiv:1812.05719, 2018.
Sparse Kalman Filtering Approaches to Realized Covariance Estimation from High Frequency Financial Data (with M. Ho), Mathematical Programming, Series B, online Feb. 14, 2019. DOI:https://doi.org/10.1007/s10107-019-01371-6.
Understanding Straight-Through Estimator in Training Activation Quantized Neural Nets
(with P. Yin, J. Lyu, S. Zhang, S. Osher, Y-Y. Qi),
to appear in Proceedings of International Conference on Learning Representations (ICLR), 2019.
Blended Coarse Gradient Descent for Full Quantization of Deep Neural Networks (with P. Yin, S. Zhang, J. Lyu, S. Osher, Y-Y. Qi), Research in the Mathematical Sciences,
DOI:10.1007/s40687-018-0177-6, to appear;
arXiv preprint arXiv:1808.05240, 2018.
BinaryRelax: A Relaxation Approach for Training Deep Neural Networks with Quantized Weights (with P. Yin, S. Zhang, J. Lyu, S. Osher, Y-Y. Qi), SIAM Journal on Imaging Sciences, 11(4),: 2205-2223, 2018;
DOI:10.1137/18M1166134; arXiv preprint arXiv:1801.06313, 2018.
Quantization and Training of Low Bit-Width Convolutional Neural Networks for Object Detection (with P. Yin, S. Zhang, Y-Y. Qi), arXiv:1612.06052v2; J. Computational Mathematics, 37(3): 349-360, 2019.
Online 9/2018, Doi:10.4208/jcm.1803-m2017-0301.
Deep Learning for Real-Time Crime Forecasting and its Ternarization (with B. Wang, P. Yin, A. Bertozzi, J. Brantingham, S. Osher), arxiv:1711.08833, 2017; Chinese Annals of Mathematics, CAM Series B, to appear.
Minimization of Transformed L1 Penalty: Theory, Difference of Convex Function Algorithm, and Robust Application in Compressed Sensing (with. S. Zhang), Mathematical Programming, Series B, 169(1), pp. 307-336, 2018. online Mar. 5, 2018; https://doi.org/10.1007/s10107-018-1236-x. Springer Nature Sharing: http://rdcu.be/IoQX
Three L1 based nonconvex methods in constructing
sparse mean reverting portfolios (with X. Long, K. Solna),
J. Sci. Computing, 75(2), pp. 1156-1186, 2018. online Oct. 20, 2017, DOI:10.1007/s10915-017-0578-5.
L1 Minimization Method for
Link Flow Correction
(with P. Yin, Z. Sun, W. Jin),
Transportation Research, Part B: Methodological, 104 (2017),
pp. 398--408..
Linear Feature Transform and Enhancement of Classification on Deep Neural Network (with P. Yin, Y-Y. Qi),
CAM report 16-41, UCLA; J. Sci. Computing, 76(3), pp. 1396-1406, 2018.
Online Feb. 23, 2018, https://doi.org/10.1007/s10915-018-0666-1
Difference-of-Convex Learning: Directional Stationarity, Optimality,
and Sparsity (with M. Ahn, J-S. Pang), SIAM J. Optimization, 27(3), pp. 1637-1665, 2017.
Transformed Schatten-1 Iterative Thresholding Algorithms for
Low Rank Matrix Completion (with S. Zhang, P. Yin), Comm. Math Sci, 15(3),
pp. 839--862, 2017.
Computational Modeling of
Spectral Data Fitting with
Nonlinear Distortions
(with Y. Sun, W. Wu),
Signal, Image, and Video Processing, 11(4):651-658 (2017).
DOI 10.1007/s11760-016-1006-2.
Iterative L1 minimization for non-convex compressed sensing (with P. Yin),
J. Computational Mathematics, 35(4), 2017, pp. 437--449.
Minimization of transformed L1 penalty: closed form representation and iterative thresholding algorithms (with S. Zhang), Comm. Math Sci,
15(2), pp. 511--537, 2017.
Point Source Super-resolution via Non-convex L1 Based
Methods (with Y. Lou, P. Yin),
J. Sci. Computing, 68(3), pp. 1082-1100, 2016.
DOI
10.1007/s10915-016-0169-x
Weighted Elastic Net Penalized Mean-Variance Portfolio Design and Computation
(with M. Ho, Z. Sun), SIAM J. Financial Mathematics, Vol. 6,
pp. 1220-1244, 2015.
A Weighted Difference of Anisotropic and Isotropic Total Variation Model for Image Processing
(with. Y. Lou, T. Zeng, S. Osher),
SIAM J. Imaging Sci, 8(3), pp. 1798-1823, 2015.
Parallelization of a Color-Entropy
Preprocessed
Chan-Vese Model for Face Contour Detection on
Multi-core CPU and GPU
(with. X. Shi, F. Park. L. Wang, Y-Y. Qi),
Parallel Computing, 49(2015), pp. 28-49,
http://dx.doi.org/10.1016/j.parco.2015.07.002. CAM Report, 15-43, UCLA.
PhaseLiftOff: an Accurate and Stable Phase Retrieval Method Based on Difference of Trace and Frobenius Norms (with P. Yin),
Comm. Math Sciences, 13(4), pp. 1033-1049, 2015.
Computational aspects of constrained L1-L2 minimization
for compressive sensing (with Y. Lou, S. Osher),
in Modelling, Computation and Optimization in Information Systems and
Management Sciences, Springer Series on
Advances in Intelligent Systems and Computing 359,
pp. 169--180, eds. H.A. Le Thi, T. Pham Dinh, N.T. Nguyen, 2015.
CAM Report 15-08, UCLA.
Minimization of L1-L2 for Compressed Sensing (with P. Yin, Y. Lou, Q. He),
SIAM J. Scientific Computing, 37(1), pp. A536-A563, 2015.
Computing Sparse Representation in a
Highly Coherent Dictionary Based on Difference of L1 and L2 (with Y. Lou,
P. Yin, Q. He),
J. Scientific Computing, Volume 64, Issue 1 (2015), pp. 178-196.
Online: Oct 2014, DOI 10.1007/s10915-014-9930-1.
A Geometric Blind Source Separation Method Based on
Facet Component Analysis (with P. Yin, Y. Sun),
Signal, Image, and Video Processing, 10(1), pp. 19-28, 2016.
DOI 10.1007/s11760-014-0696-6).
A sparse semi-blind source identification method and its application to Raman spectroscopy for explosives detection (with Y. Sun),
Signal Processing, Vol. 96, Part B, pp. 332-345, 2014.
Partially blind deblurring of barcode from out-of-focus blur
(with Y. Lou, E. Esser, H.K. Zhao),
SIAM Journal on Imaging Sciences, 7(2), 2014, pp. 740--760.
A Semi-Blind Source Separation Method for
Differential Optical Absorption Spectroscopy of
Atmospheric Gas Mixtures (with Y. Sun, L.M. Wingen,
B.J. Finlayson-Pitts), Inverse Problems and Imaging, 8(2), 2014, pp. 587-610.
Ratio and Difference of L1 and L2 Norms and Sparse
Representation with Coherent Dictionaries
(with P. Yin, E. Esser),
Communications in Information & Systems,
14(2), 2014, pp. 87--109.
A Method for Finding Structured Sparse Solutions
to Non-negative Least Squares Problems
with Applications (with E. Esser, Y. Lou),
SIAM Journal on Imaging Sciences, 6(4), pp. 2010--2046, 2013.
A Randomly Perturbed Infomax Algorithm for Blind Source Separation (with Q. He),Proceedings of the 38th International Conference on Acoustics, Speech,
and Signal Processing (ICASSP), Vancouver, pp. 3218 - 3222, May 2013.
Hybrid Deterministic-Stochastic Gradient Langevin
Dynamics for Bayesian Learning
(with Q. He),
Communications in Information & Systems
Vol. 12, No. 3, pp. 221-232, 2012.
Nonnegative Sparse Blind Source Separation for NMR Spectroscopy by Data Clustering, Model Reduction,
and L1 Minimization (with Y. Sun), SIAM J. Imaging Sciences, 5(3), 2012, pp. 886-911.
A Convex Model for Non-negative Matrix Factorization and
Dimensionality Reduction
on Physical Space
(with E. Esser, M. Moller, S. Osher, G. Sapiro),
IEEE Transactions on Image Processing, 21(7), pp. 3239-3252, 2012.
Multi-Channel L1 Regularized Convex Speech Enhancement Model and
Fast Computation by the Split Bregman Method (with M. Yu, W. Ma, S. Osher),
IEEE Transactions on Audio, Speech and Language Processing,
20(2), pp 661-675, 2012. DOI:10.1109/TASL.2011.2164526.
A Recursive Sparse Blind Source Separation Method and its Application to Correlated Data
in NMR Spectroscopy of Biofluids (with Y. Sun), Journal of Scientific Computing, 51(3), 2012,
pp. 733--753.
A Convex Model and L1 Minimization for Musical Noise Reduction in Blind Source Separation (with W. Ma, M. Yu, S. Osher), Comm Math Sciences, 10(1), pp 223-238, 2012.
Under-determined Sparse Blind Source Separation of Nonnegative and Partially
Overlapped Data (with Y. Sun),
SIAM J. Scientific Computing, Vol. 33, No. 4, pp 2063-2094, 2011.
Content Adaptive Image Matching by Color-Entropy Segmentation and Inpainting
(with Y. Sun),
the 14th International Conference on Computer Analysis
of Images and Patterns, Pedro Real et al (Eds.), CAIP 2011,
LNCS 6855, pp. 471-478, 2011, Springer-Verlag.
Modeling Category Identification Using Sparse Instance Representation (with S. Zhang, M. Lee, M. Yu),
Proceedings of the 33rd Annual Conference of the Cognitive Science Society, Austin, TX, pp. 2574--2579, 2011.
Postprocessing and Sparse Blind Source Separation of Positive and Partially
Overlapped Data (with Y. Sun, C. Ridge, F. del Rio, A.J. Shaka),
Signal Processing 91(8)(2011), pp 1838-1851.
Convexity and Fast Speech Extraction by Split Bregman Method,
(with M. Yu, W. Ma, S. Osher), Interspeech 2010, pp 398-401, Sept 26-30,
2010, Chiba, Japan.
Reducing Musical Noise in Blind Source Separation by Time-Domain
Sparse Filters and Split Bregman Method (with W. Ma, M. Yu, S. Osher),
Interspeech 2010, pp 402-405, Sept 26-30, 2010,
Chiba, Japan.
Stochastic Approximation and a Nonlocally Weighted Soft-Constrained Recursive
Algorithm for Blind Separation of Reverberant Speech Mixtures (with M. Yu),
Discrete and Continuous Dynamical Systems-A, Vol 28, No. 4, pp 1753-1767, 2010.
A Nonlocally Weighted Soft-Constrained Natural Gradient Algorithm for Blind
Source Separation of Reverberant Speech
(with M. Yu, Y-Y Qi, H-I Yang, F-G Zeng), Proceedings of IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA), pp 81-84, Oct 2009, New Paltz, NY, eds. Jacob Benesty and Tomas Gaensler.
A Time Domain Blind Decorrelation Method of Convolutive Mixtures Based on an IIR Model
(with J. Liu, Y-Y Qi), J. Computational Mathematics, Vol. 28, No. 3, pp 371-385, 2010.
A Soft-Constrained Dynamic Iterative Method of Blind Source Separation
(with J. Liu, Y-Y Qi), SIAM J. Multiscale Modeling Simulations, Vol. 7, No. 4,
pp 1795-1810, 2009.
A Time Domain Algorithm for Blind
Separation of Convolutive Sound Mixtures
and L-1 Constrained Minimization of
Cross Correlations
(with J. Liu, Y-Y Qi, F-G. Zeng),
Comm. Math Sci, Vol. 7, No. 1, 2009, pp 109-128.
A Dynamic Algorithm for Blind Separation of Convolutive Sound Mixtures
(with J. Liu, Y-Y Qi),
Neurocomputing, 72(2008), pp 521-532.
A Many to One Discrete Auditory Transform
(with Y-Y Qi),
Proceedings of ICCM 2007, Vol. III, pp 812-826, Hangzhou, China.
Ear Modeling and Sound Signal Processing,
Proceedings of ICCM 2004, Hong Kong; New Studies in Advanced Mathematics, ed. S-T Yau,
Vol. 42, pp 819 -- 830, AMS and International Press, 2008.
A Study of Hearing Aid Gain Functions Based on a Nonlinear Nonlocal
Feedforward Cochlea Model (with Y.S. Kim, Y-Y Qi),
Hearing Research, Volume 215, Issues 1-2, May 2006, pp 84-96.
Signal Processing of Acoustic Signals in the Time Domain with an Active
Nonlinear Nonlocal Cochlear Model (with M. D. LaMar, Y-Y Qi),
Signal Processing, 86(2006), pp 360-374.
An Orthogonal Discrete Auditory Transform (with Y-Y Qi),
Communications in Math Sciences, Vol. 3, No. 2, pp 251-259, 2005.
A two-dimensional nonlinear nonlocal
feed-forward cochlear model and time domain computation of
multitone interactions (with Y.S. Kim), SIAM J. Multiscale Modeling
and Simulation, Vol. 4, No. 2, pp 664-690, 2005.
An Invertible Discrete Auditory Transform (with Y-Y Qi),
Communications in Math Sciences, Vol. 3, No. 1, pp 47-56, 2005.
Dispersive instability and its minimization in time
domain computation of steady state responses of cochlear models,
Journal of the Acoustical Society of America, 115 (5), Pt. 1,
2173-2177, 2004.
Global well-posedness and multi-tone solutions
of a class of nonlinear nonlocal cochlear models in hearing (with Y-Y Qi),
Nonlinearity, Vol. 17 (2004), No. 2, pp 711-728.
A PDE based two level model of the masking property of the human ear (with Y-Y Qi),
Communications in Math Sciences, 2003, Vol. 1, No. 4, pp. 833-840.
Time Domain Computation of a Nonlinear Nonlocal Cochlear Model
with Applications to Multitone Interaction in Hearing (with Y-Y Qi, L. Deng),
Communications in Math Sciences, Vol. 1, No. 2, pp. 211-227, 2003.
Modeling Vocal Fold Motion with a Hydrodynamic Semi-Continuum
Model (with M.D. LaMar, Y-Y Qi), Journal of the Acoustical
Society of America, Vol. 114, No. 1,
pp 455-464, 2003.
A Perception and PDE Based Nonlinear Transformation
for Processing Spoken Words, (with Y-Y. Qi), Physica D 149 (2001),143-160.
Modeling Vocal Fold Motion with a Continuum Fluid Dynamic
Model (with J. M. Hyman, and Y-Y Qi),
http://arXiv.org/abs/nlin.PS/0108039.
Click here for demo.