Software
- Approximation Algorithms for D-optimal Design [codes]
We design greedy and local search for solving D-optimal design problems
The software is scalable and can be used for large-scale instances
References:
(a) Singh, M., Xie, W. (2020). Approximation Algorithms for D-optimal Design. Mathematics of Operations Research, 45(4), 1193-1620. (Authors in alphabetical order) [preprint][slides]
(b) Madan, V., Singh, M., Tantipongpipat, U., Xie, W. (2019). Combinatorial Algorithms for Optimal Design. In COLT 2019: Conference on Learning Theory (pp. 2210-2258). (Authors in alphabetical order) [paper]
We design an iterative refining strategy (IRS) to solve the large-scale instances of fair classification
The software will both improve the classification accuracy and conduct the unbiased subdata selection in an alternating fashion
It applies to fair SVM, fair logistic regression, fair CNNs
Reference: Ye, Q.*, Xie, W. (2020). Unbiased Subdata Selection for Fair Classification: A Unified Framework and Scalable Algorithms. Submitted. [preprint][slides]
We develop and generalize the ALSO-X algorithm, originally proposed by Ahmed, Luedtke, SOng, and Xie (2017), for solving a chance-constrained program
We improve ALSO-X with an alternating minimization subroutine, termed "ALSO-X+" algorithm
The software will improve the accuracy of the well-known convex approximation, CVaR method
Reference: Jiang, N., Xie, W. (2020). ALSO-X is Better Than CVaR: Convex Approximations for Chance Constrained Programs Revisited. Submitted. [preprint][slides]
We design and analyze the local search algorithm to solve the maximum entropy sampling problem with very large-scale instances
Our algorithm can be numerically demonstrated to yield less than 1% optimality gap
Reference: Li, Y.*, Xie, W. (2020). Best Principal Submatrix Selection for the Maximum Entropy Sampling Problem: Scalable Algorithms and Performance Guarantees. Submitted. [preprint][poster][slides][video]
We design Cluster-aware Supervised Learning (CluSL) frameworks and algorithms to explore clustering structures and improve learning results in the supervised learning
The software will be applied to cluster-wise regression, cluster-wise classification, cluster-wise CNNs
It improves the conventional packages such as random forests, SVC, CNNs
Reference: Chen, S., Xie, W. (2020). On the Cluster-aware Supervised Learning (CluSL): Frameworks, Convergent Algorithms, and Applications. INFORMS Journal on Computing. Accepted. [preprint][slides]
We design a fast implementation for the greedy (i.e., forward selection) method on solving the sparse ridge regression
The software can be 10 times faster than the state-of-art
Reference: Xie, W., Deng, X. (2020). Scalable Algorithms for the Sparse Ridge Regression. SIAM Journal on Optimization, 30(4), 3359–3386. [preprint][slides]