arXiv research

Non-linear kernel methods can be approximated by fast linear ones using suitable explicit feature maps allowing their application to large scale problems. We investigate how convolution kernels for structured data are composed from base kernels and construct corresponding feature maps. On this basis we propose exact an…

Nonlinear kernels can be approximated using finite-dimensional feature maps for efficient risk minimization. Due to the inherent trade-off between the dimension of the (mapped) feature space and the approximation accuracy, the key problem is to identify promising (explicit) features leading to a satisfactory out-of-sam…

Neural networks can separate non-separable data using feature maps.

The paper studies Lipschitz bounds for integral kernels under differentiability assumptions.

Proposes a new K-means method for efficient clustering of nonlinear data.

Deep neural-kernel models combine neural networks and kernel machines for scalable large datasets.

Approximating non-linear kernels using feature maps has gained a lot of interest in recent years due to applications in reducing training and testing times of SVM classifiers and other kernel based learning algorithms. We extend this line of work and present low distortion embeddings for dot product kernels into linear…

Recent years have demonstrated that using random feature maps can significantly decrease the training and testing times of kernel-based algorithms without significantly lowering their accuracy. Regrettably, because random features are target-agnostic, typically thousands of such features are necessary to achieve accept…

Kernel approximation methods create explicit, low-dimensional kernel feature maps to deal with the high computational and memory complexity of standard techniques. This work studies a supervised kernel learning methodology to optimize such mappings. We utilize the Discriminant Information criterion, a measure of class …

The Nyström methods have been popular techniques for scalable kernel based learning. They approximate explicit, low-dimensional feature mappings for kernel functions from the pairwise comparisons with the training data. However, Nyström methods are generally applied without the supervision provided by the training labe…

We present effective methods to compute equivariant harmonic maps from the universal cover of a surface into a nonpositively curved space. By discretizing the theory appropriately, we show that the energy functional is strongly convex and derive convergence of the discrete heat flow to the energy minimizer, with explic…

We consider the problem of improving the efficiency of randomized Fourier feature maps to accelerate training and testing speed of kernel methods on large datasets. These approximate feature maps arise as Monte Carlo approximations to integral representations of shift-invariant kernel functions (e.g., Gaussian kernel).…

Explicit presentations found for asymptotically rigid mapping class groups.

A new method integrates Fourier basis expansion and mapping for improved time series forecasting.

In this article we address a number of features of the moduli space of spherical metrics on connected, compact, orientable surfaces with conical singularities of assigned angles, such as its non-emptiness and connectedness. We also consider some features of the forgetful map from the above moduli space of spherical sur…

KPCA improves OoD detection by separating InD and OoD data.

Develops methods to construct harmonic and wave maps into variable-curvature surfaces.

We use filtrations of the Grassmannian model to produce explicit algebraic formulae for all harmonic maps of finite uniton number from a Riemann surface, and so all harmonic maps from the 2-sphere, to the unitary group for a general class of factorizations by unitons. We show how these specialize to give explicit formu…

This paper introduces a novel framework for generative models based on Restricted Kernel Machines (RKMs) with joint multi-view generation and uncorrelated feature learning, called Gen-RKM. To enable joint multi-view generation, this mechanism uses a shared representation of data from various views. Furthermore, the mod…

Kernel methods form a theoretically-grounded, powerful and versatile framework to solve nonlinear problems in signal processing and machine learning. The standard approach relies on the \emph{kernel trick} to perform pairwise evaluations of a kernel function, leading to scalability issues for large datasets due to its …

Explicit high-order feature interactions efficiently capture essential structural knowledge about the data of interest and have been used for constructing generative models. We present a supervised discriminative High-Order Parametric Embedding (HOPE) approach to data visualization and compression. Compared to deep emb…

We provide an explicit section for a mapping class group sequence.

New real algebraic maps with prescribed images and compositions are constructed locally like moment maps.

Identifying small subsets of features that are relevant for prediction and/or classification tasks is a central problem in machine learning and statistics. The feature selection task is especially important, and computationally difficult, for modern datasets where the number of features can be comparable to, or even ex…

Kernel methods are an incredibly popular technique for extending linear models to non-linear problems via a mapping to an implicit, high-dimensional feature space. While kernel methods are computationally cheaper than an explicit feature mapping, they are still subject to cubic cost on the number of points. Given only …

Maps persistence diagrams into Hilbert and Euclidean spaces with explicit distortions.

Deep learning maps tongue movements to speech sounds for voiceless individuals.

UniPhyNet improves cognitive load classification accuracy using EEG, ECG, and EDA signals.

Forest tree species mapped with high accuracy using satellite data.

Kernel methods form a powerful, versatile, and theoretically-grounded unifying framework to solve nonlinear problems in signal processing and machine learning. The standard approach relies on the kernel trick to perform pairwise evaluations of a kernel function, which leads to scalability issues for large datasets due …

Unified representation for tree ensembles indexed by nodes

Researchers found uncountable harmonic self-maps in complex projective spaces.

Research aims to ensure fair classification across explicit and implicit sensitive features.

A new embedding method for high-dimensional data.

Paper proposes using pairwise feature comparisons to infer modification costs for user recourse.

The study restricts manifolds with certain explicit SGL maps and constructs them.

Variant of previous work on smooth algebraic functions with compact and non-compact preimages.

Random feature maps improve forecasting with cheaper computation.

Improves generative model coverage of underrepresented modes.

We construct explicit examples of Dirac-harmonic maps $(φ, ψ)$ between Riemannian manifolds $(M,g)$ and $(N,g')$ which are non-trivial in the sense that $φ$ is not harmonic. When $\dim M=2$, we also produce examples where $φ$ is harmonic, but not conformal, and $ψ$ is non-trivial.

We present novel graph kernels for graphs with node and edge labels that have ordered neighborhoods, i.e. when neighbor nodes follow an order. Graphs with ordered neighborhoods are a natural data representation for evolving graphs where edges are created over time, which induces an order. Combining convolutional subgra…

Fold maps are higher dimensional versions of Morse functions and fundamental and important tools in studying algebraic and differential topological properties of manifolds: as the theory established by Morse and the higher dimensional version, started by Thom and Whitney, later actively studied by Eliashberg, Levine et…

Quantum machine learning models can approximate any continuous function.

We survey explicit coordinate descriptions for two (A and X) versions of Teichmuller and lamination spaces for open 2D surfaces, and extend them to the more general set-up of surfaces with distinguished collections of points on the boundary. Main features, such as mapping class group action, Poisson and symplectic stru…

Kernel ridgeless regression with random features shows good generalization without explicit regularization.

Orthogonal random features approximate a Bessel kernel, offering sharper bounds than random Fourier features.

We present an explicit description of all harmonic maps of finite uniton number from a Riemann surface into a complex Grassmannian. Namely, starting from a constant map $Q$ and a collection of meromorphic functions and their derivatives, we show how to algebraically construct all harmonic maps from the two-sphere into …

Authors create stable proper biharmonic maps from unit ball to spheres.