Sharp condition found for Burer-Monteiro method to work for MaxCut-type SDPs.
problem MaxCut-type semidefinite programs with low-rank solutions.
method Sharp condition on Laplacian matrix conditioning for global minimizers of non-convex problem.
result Any second-order critical point is a global minimizer under the given condition.
Improved guarantees for nonconvex matrix factorization with rank overparameterization.
problem Minimizing nonconvex objective over low-rank matrices.
method Overparameterized Burer--Monteiro approach, leveraging smoothness and strong convexity.
result Local optimization globally converges to global optimum under certain rank conditions.
We study the projected gradient descent method on low-rank matrix problems with a strongly convex objective. We use the Burer-Monteiro factorization approach to implicitly enforce low-rankness; such factorization introduces non-convexity in the objective. We focus on constraint sets that include both positive semi-defi…
The paper reformulates clustering as matrix factorization on the Stiefel manifold.
problem Clustering high-dimensional data like images and gene expression.
method Reformulates clustering as low-rank matrix estimation, using Burer-Monteiro factorization on the Stiefel manifold.
result Proves novel prediction bounds for clustering and proposes a componentwise Langevin sampler.
Improved understanding of low-rank solutions in SDPs via smoothed analysis.
problem Finding low-rank solutions to semidefinite programs efficiently.
method Penalty function formulation and smoothed analysis to avoid worst-case matrices.
result All approximate local optima are global optima for rank-constrained SDPs under certain conditions.
New algorithm solves large SDPs with provable convergence guarantees.
problem Solving large SDPs with diagonal constraints efficiently.
method Block-coordinate maximization applied to Burer-Monteiro method.
result Global convergence to first-order stationary point with sublinear rate.
We address the rectangular matrix completion problem by lifting the unknown matrix to a positive semidefinite matrix in higher dimension, and optimizing a nonconvex objective over the semidefinite factor using a simple gradient descent scheme. With O(μr2κ2nmax(μ,logn)) random observations of a $n_1 \times n…
Paper shows Burer-Monteiro method can solve SDPs in polynomial time under smoothed analysis.
problem Solving large-scale semidefinite programs (SDPs) efficiently.
method Perturbing SDP to create a nonconvex program in Y where Y is an nimesp matrix. result The Burer-Monteiro method can solve SDPs to any desired accuracy in polynomial time under certain conditions.
We consider the non-square matrix sensing problem, under restricted isometry property (RIP) assumptions. We focus on the non-convex formulation, where any rank-r matrix X∈Rm×n is represented as UV⊤, where U∈Rm×r and V∈Rn×r. In this paper…
Unified framework guarantees exactness of asymmetric low-rank SDP learning.
problem Exactness of asymmetric low-rank SDP learning with a quadratic penalty.
method Unified regularized asymmetric Burer-Monteiro factorization framework.
result Explicit lower bound on penalty parameter γ for exactness. Gradient descent with preconditioning finds global optima in overparameterized nonconvex factorization.
problem Finding global optima in nonconvex Burer-Monteiro factorization.
method Preconditioned gradient descent for overparameterized nonconvex function minimization.
result Gradient descent with preconditioning achieves linear convergence in the overparameterized case.
New algorithm improves clustering accuracy without sacrificing scalability.
problem Improving clustering accuracy for large datasets.
method Nonnegative low-rank semidefinite programming with Burer-Monteiro factorization.
result Significantly smaller mis-clustering errors compared to existing methods.
Paper develops efficient AltMin algorithm for SRPCP robust matrix recovery.
problem SRPCP model robust matrix recovery with universal penalty parameter.
method Tuning-free alternating minimization (AltMin) algorithm with closed-form subproblems.
result Efficient AltMin algorithm confirms robustness and efficiency.
Efficiently recovers low-tubal-rank tensors from few measurements.
problem Recovering tensors with low tubal-rank from limited measurements.
method Factorization and factorized gradient descent.
result Factorized gradient descent reduces computational costs and storage requirements.
This work shows that a simple local search can recover true principal components in non-negative rank-1 RPCA.
problem Recovering true principal components in non-negative rank-1 robust principal component analysis with noisy measurements.
method Using the Burer-Monteiro approach to cast RPCA as a non-convex and non-smooth ℓ1 optimization problem. result The low-dimensional formulation of symmetric and asymmetric positive rank-1 RPCA has a unique global solution and no spurious local solutions.
A new algorithm solves semidefinite programs using Langevin diffusion.
problem Optimizing semidefinite programs with diagonal constraints.
method Langevin diffusion on a product manifold of spheres.
result Langevin algorithm achieves ε accuracy in Ω(ε^-5) iterations.
Paper develops methods for non-quadratic loss low-rank matrix recovery.
problem Recovery of low-rank matrices with non-quadratic losses.
method Projected gradient method with a regularity projection oracle.
result Projected gradient method converges globally and linearly.
APGD algorithm reconstructs point set from partial distance measurements.
problem Reconstructing point set configuration from partial Euclidean distance measurements.
method Asymmetric Projected Gradient Descent (APGD) for EDMC problem.
result Global convergence and exact recovery with O(μ2r3κ2nlogn) observations. Paper tackles robust matrix completion with heavy-tailed noise.
problem Estimating a low-rank matrix from noisy incomplete data.
method Adaptive Huber loss for robustness, nonconvex algorithm with spectral initialization.
result Achieves minimax-optimal statistical estimation error under bounded second moment condition.
Geometric technique determines exactness of SDP robustness certificate.
problem Certifying robustness of neural networks to adversarial examples.
method Geometric projection onto hyperbola, SDP relaxation of ReLU activation.
result SDP certificate is exact for a single hidden layer under mild assumptions.
Paper improves understanding of noisy matrix completion using convex relaxation and nonconvex optimization.
problem Estimating a low-rank matrix from noisy partial entries.
method Combining convex relaxation and the nonconvex Burer-Monteiro approach.
result Convex relaxation achieves near-optimal estimation errors for noisy matrix completion.
Improved SDP relaxations for MAP inference in graphical models.
problem Efficient inference in graphical models with combinatorial optimization.
method Binary SDP relaxations using the SOS hierarchy with Burer-Monteiro method and sequential rounding.
result Demonstrated scalability to tens of thousands of variables with minutes of computation.
Simplifies solving noisy SDPs for low rank matrix recovery problems.
problem Solving SDPs with noisy data for low rank matrix recovery problems.
method Identifies conditions called simplicity to limit error in noisy SDP solutions.
result Simple SDPs can be efficiently solved and their approximate solutions trusted.
Estimates smooth function modulo 1 samples robustly from noisy data.
problem Estimating smooth function modulo 1 samples from noisy mod 1 samples.
method Formulates and solves a smoothness regularized least-squares problem over the unit circle.
result Proves robustness to noise for adversarial, Gaussian, and Bernoulli noise models.
Develops a deep multi-factor model for factor investing with clear financial insights.
problem Lack of interpretability and unclear financial insights in non-linear factor models.
method Industry and market neutralization modules, graph attention modules, factor-attention module.
result Demonstrates effectiveness in factor investing with real-world stock market data.
Factor Engine simplifies financial factor computation and analysis in Python.
problem Efficient computation and analysis of financial factors.
method Modular, extensible Python library with decorators, integrates with data science ecosystem.
result Mispricing factors computed by Factor Engine and Stata implementation are highly similar.
New risk factors improve stress testing accuracy.
problem Improving stress testing accuracy with new risk factors.
method Adapted PCA and autoencoders for dimension reduction and interpretation.
result Aggregated risk factors enhance stress testing outcomes.
New criterion ensures recovery of latent factors in NMF with mild conditions.
problem Identifying latent factors in nonnegative matrix factorization (NMF) under mild conditions.
method Proposed a new identification criterion based on the scatteredness of one factor's rows in the nonnegative orthant.
result Latent factors can be provably identified from the NMF model with minimal structural assumptions.
New statistical factors improve portfolio risk estimation.
problem Improving estimation of portfolio risk using new statistical factors.
method Matrix factor models and statistical methods (partial F test, double selection LASSO).
result New statistical factors add explanatory power in asset pricing.
Introduces factor risk measures to assess risk relative to multiple factors.
problem Measuring risk relative to multiple factors.
method Introduces a double-argument mapping as a risk measure to assess risk relative to a vector of factors.
result Characterizes various types of factor risk measures including distortion, quantile, linear, and coherent measures.
AlphaLogics mines market logic to generate interpretable alpha factors.
problem Complex, opaque alpha factors from factor mining overlook market logic.
method Market Logic Mining, Factor Generation and Optimization, Market Logic Generation and Optimization.
result AlphaLogics improves predictive metrics and risk-adjusted returns over baselines.
Method learns shared and specific factors in multi-study gene expression data.
problem Understanding shared and specific factors in high-dimensional multi-study data.
method Nonlinear multi-study factor model with sparse variational autoencoder.
result Method recovers meaningful shared and specific factors in platelet gene expression data.
FactorGCL uses hypergraph learning to predict stock returns by mining hidden factors.
problem Mining effective factors in data-driven models is challenging due to low signal-to-noise ratio in market data.
method FactorGCL employs a hypergraph structure and temporal residual contrastive learning to extract hidden factors.
result FactorGCL outperforms existing methods and mines effective hidden factors for predicting stock returns.
We propose a nonparametric Bayesian factor regression model that accounts for uncertainty in the number of factors, and the relationship between factors. To accomplish this, we propose a sparse variant of the Indian Buffet Process and couple this with a hierarchical model over factors, based on Kingman's coalescent. We…
We present a novel factor analysis method that can be applied to the discovery of common factors shared among trajectories in multivariate time series data. These factors satisfy a precedence-ordering property: certain factors are recruited only after some other factors are activated. Precedence-ordering arise in appli…
The paper derives a formula for factorizing categorical data to improve Bayes classifiers.
problem Improving the accuracy of Bayes classifiers by effectively factoring multidimensional data.
method Derives an explicit formula for calculating the marginal likelihood of a factorized categorical dataset.
result The derived formula can be used to select the best factorization for constructing a Bayes classifier.
We introduce Bayesian multi-tensor factorization, a model that is the first Bayesian formulation for joint factorization of multiple matrices and tensors. The research problem generalizes the joint matrix-tensor factorization problem to arbitrary sets of tensors of any depth, including matrices, can be interpreted as u…
We propose a framework for constructing factor models for alpha streams. Our motivation is threefold. 1) When the number of alphas is large, the sample covariance matrix is singular. 2) Its out-of-sample stability is challenging. 3) Optimization of investment allocation into alpha streams can be tractable for a factor …
A new model Weighted-SVD improves recommendation accuracy by adjusting latent factor weights.
problem Current Matrix Factorization models assume equal weights for all latent factors, which may not be accurate.
method Integrates linear regression with SVD to allow different weights for latent factors.
result The Weighted-SVD model outperforms other models in RMSE metrics on multiple datasets.
A study finds that only a few factors explain corporate bond risk, rendering extensive bond factor literature redundant.
problem The redundancy of extensive bond factor literature in explaining corporate bond risk premia.
method Bayesian Model Averaging Stochastic Discount Factor analysis of 18 quadrillion models.
result A Bayesian Model Averaging SDF explains risk premia better than low-dimensional models, with an out-of-sample Sharpe ratio of 1.5 to 1.8.
Large language models improve futures market factor models in China.
problem Designing effective factor models for Chinese futures markets.
method Used large language models (GPT) to generate 40 factors for single and multi-factor portfolios.
result GPT-generated factors outperform benchmarks with high Sharpe ratios and alphas.
New tests for identifying the number of latent factors in short panels with small time dimensions.
problem Determining the number of latent factors in short panels with small time dimensions.
method Eigenvalue tests based on variance-covariance matrices of asset returns, with assumptions on spherical errors or instrumental variables for factor betas.
result Established asymptotic distributional results and proposed a novel statistical test for weak factors.
Optimal tensor PCA for estimating factors and loadings in high-dimensional panel data.
problem Estimating factors and loadings in high-dimensional panel data with non-negligible correlations.
method Tensor Principal Component Analysis (TPCA) for estimating factors and loadings in a tensor factor model.
result Simple TPCA is optimal for strong factors and can be improved for weak factors with alternating least-squares iterations.
Paper proposes NNAFC for automatic financial factor construction.
problem Manual factor construction is time-consuming and prone to bias.
method NNAFC uses neural networks to automatically construct diversified financial factors.
result NNAFC outperforms GP in constructing more informative and diversified factors.
We give a simple explicit algorithm for building multi-factor risk models. It dramatically reduces the number of or altogether eliminates the risk factors for which the factor covariance matrix needs to be computed. This is achieved via a nested "Russian-doll" embedding: the factor covariance matrix itself is modeled v…
Model predicts video factors independently without cross-factor interference.
problem Predicting video factors without cross-factor interference.
method Unsupervised variational model for disentangling video into independent factors.
result Often learns interpretable factors like objects in a scene.
Sparse GFA identifies disease factors in FTD subgroups.
problem Heterogeneity in neurological disorders hinders understanding and treatment.
method Sparse Group Factor Analysis (GFA) with regularised horseshoe priors.
result Identified latent disease factors differentially expressed in FTD subgroups.
AlphaForge mines and dynamically combines alpha factors for better investment performance.
problem Inconsistency and inflexibility of fixed factor weights in alpha factor mining.
method Generative-predictive neural network for factor generation and dynamic weight adjustment.
result Demonstrated superior performance in formulaic alpha factor mining and portfolio returns.