arXiv subjects

Kenneth L. Clarkson

Publications and source records attributed to Kenneth L. Clarkson.

29 records · Page 2Linked to original sources

arXiv2018

Data Infrastructure and Approaches for Ontology-Based Drug Repurposing

We report development of a data infrastructure for drug repurposing that takes advantage of two currently available chemical ontologies. The data infrastructure includes a database of compound- target associations augmented with molecular ontological labels. It also contains two computational tools for prediction of new associations. We describe two drug-repurposing systems: one, Nascent Ontological Information Retrieval for Drug Repurposing (NOIR-DR), based on an information retrieval strategy, and another, based on non-negative matrix factorization together with compound similarity, that was inspired by recommender systems. We report the performance of both tools on a drug-repurposing task.

cs.IR↗

arXiv2017

Faster Kernel Ridge Regression Using Sketching and Preconditioning

Kernel Ridge Regression (KRR) is a simple yet powerful technique for non-parametric regression whose computation amounts to solving a linear system. This system is usually dense and highly ill-conditioned. In addition, the dimensions of the matrix are the same as the number of data points, so direct methods are unrealistic for large-scale datasets. In this paper, we propose a preconditioning technique for accelerating the solution of the aforementioned linear system. The preconditioner is based on random feature maps, such as random Fourier features, which have recently emerged as a powerful technique for speeding up and scaling the training of kernel-based methods, such as kernel ridge regression, by resorting to approximations. However, random feature maps only provide crude approximations to the kernel function, so delivering state-of-the-art results by directly solving the approximated system requires the number of random features to be very large. We show that random feature maps can be much more effective in forming preconditioners, since under certain conditions a not-too-large number of random features is sufficient to yield an effective preconditioner. We empirically evaluate our method and show it is highly effective for datasets of up to one million training examples.

math.NA↗

arXiv2017

Sharper Bounds for Regularized Data Fitting

We study matrix sketching methods for regularized variants of linear regression, low rank approximation, and canonical correlation analysis. Our main focus is on sketching techniques which preserve the objective function value for regularized problems, which is an area that has remained largely unexplored. We study regularization both in a fairly broad setting, and in the specific context of the popular and widely used technique of ridge regularization; for the latter, as applied to each of these problems, we show algorithmic resource bounds in which the {\em statistical dimension} appears in places where in previous bounds the rank would appear. The statistical dimension is always smaller than the rank, and decreases as the amount of regularization increases. In particular, for the ridge low-rank approximation problem $\min_{Y,X} \lVert YX - A \rVert_F^2 + λ\lVert Y\rVert_F^2 + λ\lVert X \rVert_F^2$, where $Y\in\mathbb{R}^{n\times k}$ and $X\in\mathbb{R}^{k\times d}$, we give an approximation algorithm needing \[ O(\mathtt{nnz}(A)) + \tilde{O}((n+d)\varepsilon^{-1}k \min\{k, \varepsilon^{-1}\mathtt{sd}_λ(Y^*)\})+ \mathtt{poly}(\mathtt{sd}_λ(Y^*) \varepsilon^{-1}) \] time, where $s_λ(Y^*)\le k$ is the statistical dimension of $Y^*$, $Y^*$ is an optimal $Y$, $\varepsilon$ is an error parameter, and $\mathtt{nnz}(A)$ is the number of nonzero entries of $A$.This is faster than prior work, even when $λ=0$. We also study regularization in a much more general setting. For example, we obtain sketching-based algorithms for the low-rank approximation problem $\min_{X,Y} \lVert YX - A \rVert_F^2 + f(Y,X)$ where $f(\cdot,\cdot)$ is a regularizing function satisfying some very general conditions (chiefly, invariance under orthogonal transformations).

cs.DS↗

arXiv2015

Input Sparsity and Hardness for Robust Subspace Approximation

In the subspace approximation problem, we seek a k-dimensional subspace F of R^d that minimizes the sum of p-th powers of Euclidean distances to a given set of n points a_1, ..., a_n in R^d, for p >= 1. More generally than minimizing sum_i dist(a_i,F)^p,we may wish to minimize sum_i M(dist(a_i,F)) for some loss function M(), for example, M-Estimators, which include the Huber and Tukey loss functions. Such subspaces provide alternatives to the singular value decomposition (SVD), which is the p=2 case, finding such an F that minimizes the sum of squares of distances. For p in [1,2), and for typical M-Estimators, the minimizing $F$ gives a solution that is more robust to outliers than that provided by the SVD. We give several algorithmic and hardness results for these robust subspace approximation problems. We think of the n points as forming an n x d matrix A, and letting nnz(A) denote the number of non-zero entries of A. Our results hold for p in [1,2). We use poly(n) to denote n^{O(1)} as n -> infty. We obtain: (1) For minimizing sum_i dist(a_i,F)^p, we give an algorithm running in O(nnz(A) + (n+d)poly(k/eps) + exp(poly(k/eps))), (2) we show that the problem of minimizing sum_i dist(a_i, F)^p is NP-hard, even to output a (1+1/poly(d))-approximation, answering a question of Kannan and Vempala, and complementing prior results which held for p >2, (3) For loss functions for a wide class of M-Estimators, we give a problem-size reduction: for a parameter K=(log n)^{O(log k)}, our reduction takes O(nnz(A) log n + (n+d) poly(K/eps)) time to reduce the problem to a constrained version involving matrices whose dimensions are poly(K eps^{-1} log n). We also give bicriteria solutions, (4) Our techniques lead to the first O(nnz(A) + poly(d/eps)) time algorithms for (1+eps)-approximate regression for a wide class of convex M-Estimators.

cs.DS↗

arXiv2014

The Fast Cauchy Transform and Faster Robust Linear Regression

We provide fast algorithms for overconstrained $\ell_p$ regression and related problems: for an $n\times d$ input matrix $A$ and vector $b\in\mathbb{R}^n$, in $O(nd\log n)$ time we reduce the problem $\min_{x\in\mathbb{R}^d} \|Ax-b\|_p$ to the same problem with input matrix $\tilde A$ of dimension $s \times d$ and corresponding $\tilde b$ of dimension $s\times 1$. Here, $\tilde A$ and $\tilde b$ are a coreset for the problem, consisting of sampled and rescaled rows of $A$ and $b$; and $s$ is independent of $n$ and polynomial in $d$. Our results improve on the best previous algorithms when $n\gg d$, for all $p\in[1,\infty)$ except $p=2$. We also provide a suite of improved results for finding well-conditioned bases via ellipsoidal rounding, illustrating tradeoffs between running time and conditioning quality, including a one-pass conditioning algorithm for general $\ell_p$ problems. We also provide an empirical evaluation of implementations of our algorithms for $p=1$, comparing them with related algorithms. Our empirical results show that, in the asymptotic regime, the theory is a very good guide to the practical performance of these algorithms. Our algorithms use our faster constructions of well-conditioned bases for $\ell_p$ spaces and, for $p=1$, a fast subspace embedding of independent interest that we call the Fast Cauchy Transform: a distribution over matrices $Π:\mathbb{R}^n\mapsto \mathbb{R}^{O(d\log d)}$, found obliviously to $A$, that approximately preserves the $\ell_1$ norms: that is, with large probability, simultaneously for all $x$, $\|Ax\|_1 \approx \|ΠAx\|_1$, with distortion $O(d^{2+η})$, for an arbitrarily small constant $η>0$; and, moreover, $ΠA$ can be computed in $O(nd\log d)$ time. The techniques underlying our Fast Cauchy Transform include fast Johnson-Lindenstrauss transforms, low-coherence matrices, and rescaling by Cauchy random variables.

cs.DS↗

arXiv2013

Low Rank Approximation and Regression in Input Sparsity Time

We design a new distribution over $\poly(r \eps^{-1}) \times n$ matrices $S$ so that for any fixed $n \times d$ matrix $A$ of rank $r$, with probability at least 9/10, $\norm{SAx}_2 = (1 \pm \eps)\norm{Ax}_2$ simultaneously for all $x \in \mathbb{R}^d$. Such a matrix $S$ is called a \emph{subspace embedding}. Furthermore, $SA$ can be computed in $\nnz(A) + \poly(d \eps^{-1})$ time, where $\nnz(A)$ is the number of non-zero entries of $A$. This improves over all previous subspace embeddings, which required at least $Ω(nd \log d)$ time to achieve this property. We call our matrices $S$ \emph{sparse embedding matrices}. Using our sparse embedding matrices, we obtain the fastest known algorithms for $(1+\eps)$-approximation for overconstrained least-squares regression, low-rank approximation, approximating all leverage scores, and $\ell_p$-regression. The leading order term in the time complexity of our algorithms is $O(\nnz(A))$ or $O(\nnz(A)\log n)$. We optimize the low-order $\poly(d/\eps)$ terms in our running times (or for rank-$k$ approximation, the $n*\poly(k/eps)$ term), and show various tradeoffs. For instance, we also use our methods to design new preconditioners that improve the dependence on $\eps$ in least squares regression to $\log 1/\eps$. Finally, we provide preliminary experimental results which suggest that our algorithms are competitive in practice.

cs.DS↗

arXiv2012

Self-improving Algorithms for Coordinate-wise Maxima

Computing the coordinate-wise maxima of a planar point set is a classic and well-studied problem in computational geometry. We give an algorithm for this problem in the \emph{self-improving setting}. We have $n$ (unknown) independent distributions $\cD_1, \cD_2, ..., \cD_n$ of planar points. An input pointset $(p_1, p_2, ..., p_n)$ is generated by taking an independent sample $p_i$ from each $\cD_i$, so the input distribution $\cD$ is the product $\prod_i \cD_i$. A self-improving algorithm repeatedly gets input sets from the distribution $\cD$ (which is \emph{a priori} unknown) and tries to optimize its running time for $\cD$. Our algorithm uses the first few inputs to learn salient features of the distribution, and then becomes an optimal algorithm for distribution $\cD$. Let $\OPT_\cD$ denote the expected depth of an \emph{optimal} linear comparison tree computing the maxima for distribution $\cD$. Our algorithm eventually has an expected running time of $O(\text{OPT}_\cD + n)$, even though it did not know $\cD$ to begin with. Our result requires new tools to understand linear comparison trees for computing maxima. We show how to convert general linear comparison trees to very restricted versions, which can then be related to the running time of our algorithm. An interesting feature of our algorithm is an interleaved search, where the algorithm tries to determine the likeliest point to be maximal with minimal computation. This allows the running time to be truly optimal for the distribution $\cD$.

cs.CG↗

arXiv2010

Sublinear Optimization for Machine Learning

We give sublinear-time approximation algorithms for some optimization problems arising in machine learning, such as training linear classifiers and finding minimum enclosing balls. Our algorithms can be extended to some kernelized versions of these problems, such as SVDD, hard margin SVM, and L2-SVM, for which sublinear-time algorithms were not known before. These new algorithms use a combination of a novel sampling techniques and a new multiplicative update algorithm. We give lower bounds which show the running times of many of our algorithms to be nearly best possible in the unit-cost RAM model. We also give implementations of our algorithms in the semi-streaming setting, obtaining the first low pass polylogarithmic space and sublinear time algorithms achieving arbitrary approximation factor.

cs.LG↗

arXiv2010

Self-Improving Algorithms

We investigate ways in which an algorithm can improve its expected performance by fine-tuning itself automatically with respect to an unknown input distribution D. We assume here that D is of product type. More precisely, suppose that we need to process a sequence I_1, I_2, ... of inputs I = (x_1, x_2, ..., x_n) of some fixed length n, where each x_i is drawn independently from some arbitrary, unknown distribution D_i. The goal is to design an algorithm for these inputs so that eventually the expected running time will be optimal for the input distribution D = D_1 * D_2 * ... * D_n. We give such self-improving algorithms for two problems: (i) sorting a sequence of numbers and (ii) computing the Delaunay triangulation of a planar point set. Both algorithms achieve optimal expected limiting complexity. The algorithms begin with a training phase during which they collect information about the input distribution, followed by a stationary regime in which the algorithms settle to their optimized incarnations.

cs.DS↗

arXiv2009

On the Set Multi-Cover Problem in Geometric Settings

We consider the set multi-cover problem in geometric settings. Given a set of points P and a collection of geometric shapes (or sets) F, we wish to find a minimum cardinality subset of F such that each point p in P is covered by (contained in) at least d(p) sets. Here d(p) is an integer demand (requirement) for p. When the demands d(p)=1 for all p, this is the standard set cover problem. The set cover problem in geometric settings admits an approximation ratio that is better than that for the general version. In this paper, we show that similar improvements can be obtained for the multi-cover problem as well. In particular, we obtain an O(log Opt) approximation for set systems of bounded VC-dimension, where Opt is the cardinality of an optimal solution, and an O(1) approximation for covering points by half-spaces in three dimensions and for some other classes of shapes.

cs.CG↗

arXiv2005

Improved Approximation Algorithms for Geometric Set Cover

Given a collection S of subsets of some set U, and M a subset of U, the set cover problem is to find the smallest subcollection C of S such that M is a subset of the union of the sets in C. While the general problem is NP-hard to solve, even approximately, here we consider some geometric special cases, where usually U = R^d. Extending prior results, we show that approximation algorithms with provable performance exist, under a certain general condition: that for a random subset R of S and function f(), there is a decomposition of the portion of U not covered by R into an expected f(|R|) regions, each region of a particular simple form. We show that under this condition, a cover of size O(f(|C|)) can be found. Our proof involves the generalization of shallow cuttings to more general geometric situations. We obtain constant-factor approximation algorithms for covering by unit cubes in R^3, for guarding a one-dimensional terrain, and for covering by similar-sized fat triangles in R^2. We also obtain improved approximation guarantees for fat triangles, of arbitrary size, and for a class of fat objects.

cs.CG↗