Andersen Ang | Papers | Talks | Notes | Misc.

AICE2001 Scientific Computing, USouthampton, 25Fall

COMP6260 Optimisation for Machine Learning, USouthampton, 25Spring
Info | ConvexAnalysis | 1st-order algo | Other algo | Nonconvex | Lab

AICE1004 Maths for AI&CE, USouthampton, 24Fall
Info | LectureNote | MadBookPro

COMP1311 Math I, USouthampton, 24Fall
COMP1215 Foundations of Computer Science, USouthampton, 23Fall
Combinatorics Graph Probability StatisticsNote Keys & book Madbook

CO327 Deterministic OR Models, UWaterloo, 22Spring
Intro Info Review LinearProgram Tutorial1 IntegerProgram Tutorial2
FourierMotzkin Matlab1, Matlab2 BigM if-then IntegerProgram
Extreme-case trick 6Parimutuel auction -median Maxcut Piecewise-linear
2-player 0-sum gameLogicIntegerProgram OptimalTransport Example
Notes on LP & IP

MARO201 UMONS, 19-20 Gradient descent solving quadratic problems | video

• Continuous TV in infinite dimensional space

• Discrete TV in finite dimensional space

• Solving TV norm regularized least square by Proximal gradient method with ADMM

• Solving TV norm regularized least square by MM

• Submodulairty

• NNLS, its geometry & its KKT

• Penalty method is not effective for NNLS – using softmax

• Penalty method is not effective for NNLS – using iteratively reweighed least squares

• NNLS by projected gradient descent: acceleration & restart | NNLS PGD mfile | NNLS PGD mfile 2

• NNLS by multiplicative update | NNLS MU mfile

• NNLS by ADMM

• Nonnegative Unimodal Least Squares

• Step-by-step expansion of

• If A and its inverse are nonnegative, then A is the permutation of a positive diagonal matrix

• No scaling ambiguity in Symmetric NMF (hand written)

• Basic cone geometry and algebra for NMF

• AB is nonnegative if

• NMF and multiplicative update

• Convergence analysis of the multiplicative update

• NMF via projected gradient |
  PGD algorithm mfile | APG algorithm mfile

• NMF via HALS: column-wise exact block coordinate descent |
  HALS mfile

• NMF with regularization

• Multi-grid NMF

• Separable NMF: separability assumption

• Uniqueness of solution of minimum volume (N)MF

• Rank-1 NMF is not NP-Hard & closed-form sol

• KL divergence is not Lipschitz smooth

• Matrix Rank Minimization

• Introduction to Matrix Completion

• Nuclear norm is tight convex relaxation of rank function only within the unit ball

• Proving nuclear norm is a norm

• Characterizing nuclear norm: nuclear norm is the dual norm of the spectral norm

• Understanding the uniqueness of the sol. of the nuclear norm minimization

• Singular value thresholding solves

• Singular value thresholding solves by von Neumann trace inequality

• MC algorithm: Majorization-Minimization

• On the rank of matrix equation

• Matrix Factorization in vectorized format mfile

• Column-wise representation of determinant of gram matrix

• Operator norm of rank-1 perturbation

• Get rid of the Hadamard product in

• Quaternion

• Matrix derivatives, Single Entry Matrix and derivatives of logdet

• Matrix derivative on scalar function of matrix variable

• Matrix derivative on matrix function of matrix variable

• Functions of singular values

• Subgradient of functions of singular values

• Generating uniform unit random vectors in

• Johnson-Lindenstrauss Lemma & big matrices are approximately low rank

• subject to

• Uniqueness of sol. of spares recovery problem: Kruskal rank & incoherence

• Mutual Coherence & the Welch Bound

• Orthogonal Matching Pursuit

• Review numerical 2D Poisson Equation

• Elastic Obstacle Problem

• Nonlinear elastic obstacle problem

• Poisson image editing

• MGOPT direction is a descent direction