A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.
New pinball loss improves decoding of noisy one-bit compressive sensing.
DeepFPC uses neural networks to recover sparse signals from quantized measurements.
New theoretical framework improves error rates for sparse learning with convex regularization.
Feature selection is a technique to screen out less important features. Many existing supervised feature selection algorithms use redundancy and relevancy as the main criteria to select features. However, feature interaction, potentially a key characteristic in real-world problems, has not received much attention. As a…
The L1 loss landscape of neural nets near local minima behaves differently, revealing exponential decay and increased vertex density.
Improved sample efficiency in learning sparse Ising models.
Stochastic gradient descent on l1 loss converges to true parameter in online robust regression.
This paper explores neural networks for colorizing grayscale images.
This paper improves adversarial robustness of deep learning models.
We define a notion of Hempel distance for one-sided Heegaard splittings and show that the existence of alternate surfaces restricts distance for one-sided splittings in a manner similar to Hartshorn's and Scharlemann-Tomova's results for two-sided splittings. We also show that every geometrically compressible one-sided…
Study on estimating class probabilities using empirical risk minimization.
The use of machine-learning in neuroimaging offers new perspectives in early diagnosis and prognosis of brain diseases. Although such multivariate methods can capture complex relationships in the data, traditional approaches provide irregular (l2 penalty) or scattered (l1 penalty) predictive pattern with a very limited…
When a Dehn filled link manifold contains a geometrically incompressible one-sided surface, it is shown there is a unique boundary incompressible position that the surface can take in the link space. The proof uses a version of the sweep-out technique from two-sided Heegaard splitting theory. When applied to one-sided …
Dynamic pruning during training reduces deep network complexity without significant accuracy loss.
In this paper we prove an extrinsic one-sided curvature estimate for disks embedded in with constant mean curvature which is independent of the value of the constant mean curvature. We apply this extrinsic one-sided curvature estimate in [24] to prove to prove a weak chord arc type result for these disks…
Large-scale L1-regularized loss minimization problems arise in high-dimensional applications such as compressed sensing and high-dimensional supervised learning, including classification and regression problems. High-performance algorithms and implementations are critical to efficiently solving these problems. Building…
This manuscript provides optimization guarantees, generalization bounds, and statistical consistency results for AdaBoost variants which replace the exponential loss with the logistic and similar losses (specifically, twice differentiable convex losses which are Lipschitz and tend to zero on one side). The heart of the…
Develops a privacy-preserving IRLS algorithm for L1 minimization.
Using basic properties of one-sided Heegaard splittings, a direct proof that geometrically compressible one-sided splittings of RP^3 are stabilised is given. The argument is modelled on that used by Waldhausen to show that two-sided splittings of S^3 are standard.
New algorithms solve L1-regularized SVMs and related LPs, outperforming existing methods.
We propose a novel general algorithm LHAC that efficiently uses second-order information to train a class of large-scale l1-regularized problems. Our method executes cheap iterations while achieving fast local convergence rate by exploiting the special structure of a low-rank matrix, constructed via quasi-Newton approx…
Generalizes robust loss functions for improved performance in various tasks.
OPDA optimizes L1-regularized models with faster convergence and sparsity.
New method solves rank-1 L1-norm TUCKER2 decomposition efficiently.
We study the performance of a family of randomized parallel coordinate descent methods for minimizing the sum of a nonsmooth and separable convex functions. The problem class includes as a special case L1-regularized L1 regression and the minimization of the exponential loss ("AdaBoost problem"). We assume the input da…
We address the problem of Compressed Sensing (CS) with side information. Namely, when reconstructing a target CS signal, we assume access to a similar signal. This additional knowledge, the side information, is integrated into CS via L1-L1 and L1-L2 minimization. We then provide lower bounds on the number of measuremen…
The stability and the index of complete one-sided minimal surfaces of certain three-dimensional Riemannian manifolds with positive scalar curvature are studied.
A new algorithm optimizes L1-norm error fitting problems efficiently.
Adam-type optimizers show one-sided convergence in GAN training, not reaching critical points.
New algorithms improve L1 PCA performance.
A new l1-norm penalized algorithm for regression reduces computational cost.
The use of L1 regularisation for sparse learning has generated immense research interest, with successful application in such diverse areas as signal acquisition, image coding, genomics and collaborative filtering. While existing work highlights the many advantages of L1 methods, in this paper we find that L1 regularis…
In the closed, non-Haken, hyperbolic class of examples generated by (2p,q) Dehn fillings of Figure 8 knot space, the geometrically incompressible one-sided surfaces are identified by the filling ratio p/q and determined to be unique in all cases. When applied to one-sided Heegaard splittings, this can be used to classi…
Various structural properties are developed for non-orientable surfaces in link spaces. The Möbius band tree is described to represent genus growth of one-sided surfaces in solid tori. The structure of the Tree allows various insights into the change of genus under boundary slope, which are not possible using the exist…
This paper studies ordered weighted L1 (OWL) norm regularization for sparse estimation problems with strongly correlated variables. We prove sufficient conditions for clustering based on the correlation/colinearity of variables using the OWL norm, of which the so-called OSCAR is a particular case. Our results extend pr…
A new algorithm calculates L1-norm principal components efficiently for large datasets.
Adaptive l1-regularization controls short-selling in portfolio selection.
New toolkit classifies hazardous solvents from Raman spectra.
New method for sparse data using L1-NMF with improved sparsity control.
Study non-orientable surfaces to find loops winding around punctures.
Lass-0 finds sparse solutions to L0 regularized regression using efficient search.
Framework for adaptive online learning with various bounds.
A multilevel framework speeds up sparse optimization for inverse covariance estimation and logistic regression.
Proposes a new method for hyperspectral image dimensionality reduction.
We give a sufficient and necessary condition of the fundamental group homomorphism of a map between manifolds to induce homology equivalences. Moreover, a classification of one-sided h-cobordism of manifolds up to diffeomorphisms is obtained, based on Quillen's plus construction with Whitehead torsions.
Left orderability proven for certain 3-manifolds with specific foliations.
We theoretically investigate the convergence rate and support consistency (i.e., correctly identifying the subset of non-zero coefficients in the large sample limit) of multiple kernel learning (MKL). We focus on MKL with block-l1 regularization (inducing sparse kernel combination), block-l2 regularization (inducing un…
Study removes singularities from area-minimizing surfaces.