arXiv subjects

Mary Wootters

Publications and source records attributed to Mary Wootters.

70 records · Page 4Linked to original sources

arXiv2017

Average-radius list-recovery of random linear codes: it really ties the room together

We analyze the list-decodability, and related notions, of random linear codes. This has been studied extensively before: there are many different parameter regimes and many different variants. Previous works have used complementary styles of arguments---which each work in their own parameter regimes but not in others---and moreover have left some gaps in our understanding of the list-decodability of random linear codes. In particular, none of these arguments work well for list-recovery, a generalization of list-decoding that has been useful in a variety of settings. In this work, we present a new approach, which works across parameter regimes and further generalizes to list-recovery. Our main theorem can establish better list-decoding and list-recovery results for low-rate random linear codes over large fields; list-recovery of high-rate random linear codes; and it can recover the rate bounds of Guruswami, Hastad, and Kopparty for constant-rate random linear codes (although with large list sizes).

cs.IT↗

arXiv2016

Repairing Reed-Solomon Codes

We study the performance of Reed-Solomon (RS) codes for the \em exact repair problem \em in distributed storage. Our main result is that, in some parameter regimes, Reed-Solomon codes are optimal regenerating codes, among MDS codes with linear repair schemes. Moreover, we give a characterization of MDS codes with linear repair schemes which holds in any parameter regime, and which can be used to give non-trivial repair schemes for RS codes in other settings. More precisely, we show that for $k$-dimensional RS codes whose evaluation points are a finite field of size $n$, there are exact repair schemes with bandwidth $(n-1)\log((n-1)/(n-k))$ bits, and that this is optimal for any MDS code with a linear repair scheme. In contrast, the naive (commonly implemented) repair algorithm for this RS code has bandwidth $k\log(n)$ bits. When the entire field is used as evaluation points, the number of nodes $n$ is much larger than the number of bits per node (which is $O(\log(n))$), and so this result holds only when the degree of sub-packetization is small. However, our method applies in any parameter regime, and to illustrate this for high levels of sub-packetization we give an improved repair scheme for a specific (14,10)-RS code used in the Facebook Hadoop Analytics cluster.

cs.IT↗

arXiv2016

One-Bit Compressive Sensing of Dictionary-Sparse Signals

One-bit compressive sensing has extended the scope of sparse recovery by showing that sparse signals can be accurately reconstructed even when their linear measurements are subject to the extreme quantization scenario of binary samples---only the sign of each linear measurement is maintained. Existing results in one-bit compressive sensing rely on the assumption that the signals of interest are sparse in some fixed orthonormal basis. However, in most practical applications, signals are sparse with respect to an overcomplete dictionary, rather than a basis. There has already been a surge of activity to obtain recovery guarantees under such a generalized sparsity model in the classical compressive sensing setting. Here, we extend the one-bit framework to this important model, providing a unified theory of one-bit compressive sensing under dictionary sparsity. Specifically, we analyze several different algorithms---based on convex programming and on hard thresholding---and show that, under natural assumptions on the sensing matrix (satisfied by Gaussian matrices), these algorithms can efficiently recover analysis-dictionary-sparse signals in the one-bit model.

cs.IT↗

arXiv2015

Strategic Classification

Machine learning relies on the assumption that unseen test instances of a classification problem follow the same distribution as observed training data. However, this principle can break down when machine learning is used to make important decisions about the welfare (employment, education, health) of strategic individuals. Knowing information about the classifier, such individuals may manipulate their attributes in order to obtain a better classification outcome. As a result of this behavior---often referred to as gaming---the performance of the classifier may deteriorate sharply. Indeed, gaming is a well-known obstacle for using machine learning methods in practice; in financial policy-making, the problem is widely known as Goodhart's law. In this paper, we formalize the problem, and pursue algorithms for learning classifiers that are robust to gaming. We model classification as a sequential game between a player named "Jury" and a player named "Contestant." Jury designs a classifier, and Contestant receives an input to the classifier, which he may change at some cost. Jury's goal is to achieve high classification accuracy with respect to Contestant's original input and some underlying target classification function. Contestant's goal is to achieve a favorable classification outcome while taking into account the cost of achieving it. For a natural class of cost functions, we obtain computationally efficient learning algorithms which are near-optimal. Surprisingly, our algorithms are efficient even on concept classes that are computationally hard to learn. For general cost functions, designing an approximately optimal strategy-proof classifier, for inverse-polynomial approximation, is NP-hard.

cs.LG↗

arXiv2015

Linear-time list recovery of high-rate expander codes

We show that expander codes, when properly instantiated, are high-rate list recoverable codes with linear-time list recovery algorithms. List recoverable codes have been useful recently in constructing efficiently list-decodable codes, as well as explicit constructions of matrices for compressive sensing and group testing. Previous list recoverable codes with linear-time decoding algorithms have all had rate at most 1/2; in contrast, our codes can have rate $1 - ε$ for any $ε> 0$. We can plug our high-rate codes into a construction of Meir (2014) to obtain linear-time list recoverable codes of arbitrary rates, which approach the optimal trade-off between the number of non-trivial lists provided and the rate of the code. While list-recovery is interesting on its own, our primary motivation is applications to list-decoding. A slight strengthening of our result would implies linear-time and optimally list-decodable codes for all rates, and our work is a step in the direction of solving this important problem.

cs.IT↗

arXiv2015

Local Correctability of Expander Codes

In this work, we present the first local-decoding algorithm for expander codes. This yields a new family of constant-rate codes that can recover from a constant fraction of errors in the codeword symbols, and where any symbol of the codeword can be recovered with high probability by reading $N^ε$ symbols from the corrupted codeword, where $N$ is the block-length of the code. Expander codes, introduced by Sipser and Spielman, are formed from an expander graph $G = (V,E)$ of degree $d$, and an inner code of block-length $d$ over an alphabet $Σ$. Each edge of the expander graph is associated with a symbol in $Σ$. A string in $Σ^{E}$ will be a codeword if for each vertex in $V$, the symbols on the adjacent edges form a codeword in the inner code. We show that if the inner code has a smooth reconstruction algorithm in the noiseless setting, then the corresponding expander code has an efficient local-correction algorithm in the noisy setting. Instantiating our construction with inner codes based on finite geometries, we obtain novel locally decodable codes with rate approaching one. This provides an alternative to the multiplicity codes of Kopparty, Saraf and Yekhanin (STOC '11) and the lifted codes of Guo, Kopparty and Sudan (ITCS '13).

cs.IT↗

arXiv2014

It'll probably work out: improved list-decoding through random operations

In this work, we introduce a framework to study the effect of random operations on the combinatorial list-decodability of a code. The operations we consider correspond to row and column operations on the matrix obtained from the code by stacking the codewords together as columns. This captures many natural transformations on codes, such as puncturing, folding, and taking subcodes; we show that many such operations can improve the list-decoding properties of a code. There are two main points to this. First, our goal is to advance our (combinatorial) understanding of list-decodability, by understanding what structure (or lack thereof) is necessary to obtain it. Second, we use our more general results to obtain a few interesting corollaries for list decoding: (1) We show the existence of binary codes that are combinatorially list-decodable from $1/2-ε$ fraction of errors with optimal rate $Ω(ε^2)$ that can be encoded in linear time. (2) We show that any code with $Ω(1)$ relative distance, when randomly folded, is combinatorially list-decodable $1-ε$ fraction of errors with high probability. This formalizes the intuition for why the folding operation has been successful in obtaining codes with optimal list decoding parameters; previously, all arguments used algebraic methods and worked only with specific codes. (3) We show that any code which is list-decodable with suboptimal list sizes has many subcodes which have near-optimal list sizes, while retaining the error correcting capabilities of the original code. This generalizes recent results where subspace evasive sets have been used to reduce list sizes of codes that achieve list decoding capacity.

cs.IT↗

arXiv2014

Exponential decay of reconstruction error from binary measurements of sparse signals

Binary measurements arise naturally in a variety of statistical and engineering applications. They may be inherent to the problem---e.g., in determining the relationship between genetics and the presence or absence of a disease---or they may be a result of extreme quantization. In one-bit compressed sensing it has recently been shown that the number of one-bit measurements required for signal estimation mirrors that of unquantized compressed sensing. Indeed, $s$-sparse signals in $\mathbb{R}^n$ can be estimated (up to normalization) from $Ω(s \log (n/s))$ one-bit measurements. Nevertheless, controlling the precise accuracy of the error estimate remains an open challenge. In this paper, we focus on optimizing the decay of the error as a function of the oversampling factor $λ:= m/(s \log(n/s))$, where $m$ is the number of measurements. It is known that the error in reconstructing sparse signals from standard one-bit measurements is bounded below by $Ω(λ^{-1})$. Without adjusting the measurement procedure, reducing this polynomial error decay rate is impossible. However, we show that an adaptive choice of the thresholds used for quantization may lower the error rate to $e^{-Ω(λ)}$. This improves upon guarantees for other methods of adaptive thresholding as proposed in Sigma-Delta quantization. We develop a general recursive strategy to achieve this exponential decay and two specific polynomial-time algorithms which fall into this framework, one based on convex programming and one on hard thresholding. This work is inspired by the one-bit compressed sensing model, in which the engineer controls the measurement procedure. Nevertheless, the principle is extendable to signal reconstruction problems in a variety of binary statistical models as well as statistical estimation problems like logistic regression.

cs.IT↗

arXiv2014

Fast matrix completion without the condition number

We give the first algorithm for Matrix Completion whose running time and sample complexity is polynomial in the rank of the unknown target matrix, linear in the dimension of the matrix, and logarithmic in the condition number of the matrix. To the best of our knowledge, all previous algorithms either incurred a quadratic dependence on the condition number of the unknown matrix or a quadratic dependence on the dimension of the matrix in the running time. Our algorithm is based on a novel extension of Alternating Minimization which we show has theoretical guarantees under standard assumptions even in the presence of noise.

cs.LG↗

arXiv2014

1-Bit Matrix Completion

In this paper we develop a theory of matrix completion for the extreme case of noisy 1-bit observations. Instead of observing a subset of the real-valued entries of a matrix M, we obtain a small number of binary (1-bit) measurements generated according to a probability distribution determined by the real-valued entries of M. The central question we ask is whether or not it is possible to obtain an accurate estimate of M from this data. In general this would seem impossible, but we show that the maximum likelihood estimate under a suitable constraint returns an accurate estimate of M when ||M||_{\infty} <= α, and rank(M) <= r. If the log-likelihood is a concave function (e.g., the logistic or probit observation models), then we can obtain this maximum likelihood estimate by optimizing a convex program. In addition, we also show that if instead of recovering M we simply wish to obtain an estimate of the distribution generating the 1-bit measurements, then we can eliminate the requirement that ||M||_{\infty} <= α. For both cases, we provide lower bounds showing that these estimates are near-optimal. We conclude with a suite of experiments that both verify the implications of our theorems as well as illustrate some of the practical applications of 1-bit matrix completion. In particular, we compare our program to standard matrix completion methods on movie rating data in which users submit ratings from 1 to 5. In order to use our program, we quantize this data to a single bit, but we allow the standard matrix completion program to have access to the original ratings (from 1 to 5). Surprisingly, the approach based on binary data performs significantly better.

math.ST↗

arXiv2013

Every list-decodable code for high noise has abundant near-optimal rate puncturings

We show that any q-ary code with sufficiently good distance can be randomly punctured to obtain, with high probability, a code that is list decodable up to radius $1 - 1/q - ε$ with near-optimal rate and list sizes. Our results imply that "most" Reed-Solomon codes are list decodable beyond the Johnson bound, settling the long-standing open question of whether any Reed Solomon codes meet this criterion. More precisely, we show that a Reed-Solomon code with random evaluation points is, with high probability, list decodable up to radius $1 - ε$ with list sizes $O(1/ε)$ and rate $Ω(ε)$. As a second corollary of our argument, we obtain improved bounds on the list decodability of random linear codes over large fields. Our approach exploits techniques from high dimensional probability. Previous work used similar tools to obtain bounds on the list decodability of random linear codes, but the bounds did not scale with the size of the alphabet. In this paper, we use a chaining argument to deal with large alphabet sizes.

cs.IT↗

arXiv2013

Accurate Decoding of Pooled Sequenced Data Using Compressed Sensing

In order to overcome the limitations imposed by DNA barcoding when multiplexing a large number of samples in the current generation of high-throughput sequencing instruments, we have recently proposed a new protocol that leverages advances in combinatorial pooling design (group testing) doi:10.1371/journal.pcbi.1003010. We have also demonstrated how this new protocol would enable de novo selective sequencing and assembly of large, highly-repetitive genomes. Here we address the problem of decoding pooled sequenced data obtained from such a protocol. Our algorithm employs a synergistic combination of ideas from compressed sensing and the decoding of error-correcting codes. Experimental results on synthetic data for the rice genome and real data for the barley genome show that our novel decoding algorithm enables significantly higher quality assemblies than the previous approach.

q-bio.QM↗

arXiv2013

On the list decodability of random linear codes with large error rates

It is well known that a random q-ary code of rate Ω(ε^2) is list decodable up to radius (1 - 1/q - ε) with list sizes on the order of 1/ε^2, with probability 1 - o(1). However, until recently, a similar statement about random linear codes has until remained elusive. In a recent paper, Cheraghchi, Guruswami, and Velingker show a connection between list decodability of random linear codes and the Restricted Isometry Property from compressed sensing, and use this connection to prove that a random linear code of rate Ω(ε^2 / log^3(1/ε)) achieves the list decoding properties above, with constant probability. We improve on their result to show that in fact we may take the rate to be Ω(ε^2), which is optimal, and further that the success probability is 1 - o(1), rather than constant. As an added benefit, our proof is relatively simple. Finally, we extend our methods to more general ensembles of linear codes. As an example, we show that randomly punctured Reed-Muller codes have the same list decoding properties as the original codes, even when the rate is improved to a constant.

cs.IT↗

arXiv2012

On optimal entanglement assisted one-shot classical communication

The one-shot success probability of a noisy classical channel for transmitting one classical bit is the optimal probability with which the bit can be sent via a single use of the channel. Prevedel et al. (PRL 106, 110505 (2011)) recently showed that for a specific channel, this quantity can be increased if the parties using the channel share an entangled quantum state. We completely characterize the optimal entanglement-assisted protocols in terms of the radius of a set of operators associated with the channel. This characterization can be used to construct optimal entanglement-assisted protocols from the given classical channel and to prove the limit of such protocols. As an example, we show that the Prevedel et al. protocol is optimal for two-qubit entanglement. We also prove some simple upper bounds on the improvement that can be obtained from quantum and no-signaling correlations.

quant-ph↗

arXiv2012

New constructions of RIP matrices with fast multiplication and fewer rows

In compressed sensing, the "restricted isometry property" (RIP) is a sufficient condition for the efficient reconstruction of a nearly k-sparse vector x in C^d from m linear measurements Phi x. It is desirable for m to be small, and for Phi to support fast matrix-vector multiplication. In this work, we give a randomized construction of RIP matrices Phi in C^{m x d}, preserving the L_2 norms of all k-sparse vectors with distortion 1+eps, where the matrix-vector multiply Phi x can be computed in nearly linear time. The number of rows m is on the order of eps^{-2}klog dlog^2(klog d). Previous analyses of constructions of RIP matrices supporting fast matrix-vector multiplies, such as the sampled discrete Fourier matrix, required m to be larger by roughly a log k factor. Supporting fast matrix-vector multiplication is useful for iterative recovery algorithms which repeatedly multiply by Phi or Phi^*. Furthermore, our construction, together with a connection between RIP matrices and the Johnson-Lindenstrauss lemma in [Krahmer-Ward, SIAM. J. Math. Anal. 2011], implies fast Johnson-Lindenstrauss embeddings with asymptotically fewer rows than previously known. Our approach is a simple twist on previous constructions. Rather than choosing the rows for the embedding matrix to be rows sampled from some larger structured matrix (such as the discrete Fourier transform or a random circulant matrix), we instead choose each row of the embedding matrix to be a linear combination of a small number of rows of the original matrix, with random sign flips as coefficients. The main tool in our analysis is a recent bound for the supremum of certain types of Rademacher chaos processes in [Krahmer-Mendelson-Rauhut, arXiv:1207.0235].

cs.DS↗

arXiv2008

Configuration spaces of convex and embedded polygons in the plane

This paper studies the configuration spaces of linkages whose underlying graph is a single cycle. Assume that the edge lengths are such that there are no configurations in which all the edges lie along a line. The main results are that, modulo translations and rotations, each component of the space of convex configurations is homeomorphic to a closed Euclidean ball and each component of the space of embedded configurations is homeomorphic to a Euclidean space. This represents an elaboration on the topological information that follows from the convexification theorem of Connelly, Demaine, and Rote.

cs.CG↗