arXiv subjects

Mary Wootters

Publications and source records attributed to Mary Wootters.

At least 37 records · Page 2Linked to original sources

arXiv2023

A Characterization of Optimal-Rate Linear Homomorphic Secret Sharing Schemes, and Applications

A Homomorphic Secret Sharing (HSS) scheme is a secret-sharing scheme that shares a secret $x$ among $s$ servers, and additionally allows an output client to reconstruct some function $f(x)$, using information that can be locally computed by each server. A key parameter in HSS schemes is download rate, which quantifies how much information the output client needs to download from each server. Recent work (Fosli, Ishai, Kolobov, and Wootters, ITCS 2022) established a fundamental limitation on the download rate of linear HSS schemes for computing low-degree polynomials, and gave an example of HSS schemes that meet this limit. In this paper, we further explore optimal-rate linear HSS schemes for polynomials. Our main result is a complete characterization of such schemes, in terms of a coding-theoretic notion that we introduce, termed optimal labelweight codes. We use this characterization to answer open questions about the amortization required by HSS schemes that achieve optimal download rate. In more detail, the construction of Fosli et al. required amortization over $\ell$ instances of the problem, and only worked for particular values of $\ell$. We show that -- perhaps surprisingly -- the set of $\ell$'s for which their construction works is in fact nearly optimal, possibly leaving out only one additional value of $\ell$. We show this by using our coding-theoretic characterization to prove a necessary condition on the $\ell$'s admitting optimal-rate linear HSS schemes. We then provide a slightly improved construction of optimal-rate linear HSS schemes, where the set of allowable $\ell$'s is optimal in even more parameter settings. Moreover, based on a connection to the MDS conjecture, we conjecture that our construction is optimal for all parameter regimes.

cs.IT↗

arXiv2023

Viderman's algorithm for quantum LDPC codes

Quantum low-density parity-check (LDPC) codes, a class of quantum error correcting codes, are considered a blueprint for scalable quantum circuits. To use these codes, one needs efficient decoding algorithms. In the classical setting, there are multiple efficient decoding algorithms available, including Viderman's algorithm (Viderman, TOCT 2013). Viderman's algorithm for classical LDPC codes essentially reduces the error-correction problem to that of erasure-correction, by identifying a small envelope $L$ that is guaranteed to contain the error set. Our main result is a generalization of Viderman's algorithm to quantum LDPC codes, namely hypergraph product codes (Tillich, Zémor, IEEE T-IT, 2013). This is the first erasure-conversion algorithm that can correct up to $Ω(D)$ errors for constant-rate quantum LDPC codes, where $D$ is the distance of the code. In that sense, it is also fundamentally different from existing decoding algorithms, in particular from the small-set-flip algorithm (Leverrier, Tillich, Zémor, FOCS, 2015). Moreover, in some parameter regimes, our decoding algorithm improves on the decoding radius of existing algorithms. We note that we do not yet have linear-time erasure-decoding algorithms for quantum LDPC codes, and thus the final running time of the whole decoding algorithm is not linear; however, we view our linear-time envelope-finding algorithm as an important first step.

quant-ph↗

arXiv2023

Tight Limits on Nonlocality from Nontrivial Communication Complexity; a.k.a. Reliable Computation with Asymmetric Gate Noise

It has long been known that the existence of certain superquantum nonlocal correlations would cause communication complexity to collapse. The absurdity of a world in which any nonlocal binary function could be evaluated with a constant amount of communication in turn provides a tantalizing way to distinguish quantum mechanics from incorrect theories of physics; the statement "communication complexity is nontrivial" has even been conjectured to be a concise information-theoretic axiom for characterizing quantum mechanics. We directly address the viability of that perspective with two results. First, we exhibit a nonlocal game such that communication complexity collapses in any physical theory whose maximal winning probability exceeds the quantum value. Second, we consider the venerable CHSH game that initiated this line of inquiry. In that case, the quantum value is about 0.85 but it is known that a winning probability of approximately 0.91 would collapse communication complexity. We provide evidence that the 0.91 result is the best possible using a large class of proof strategies, suggesting that the communication complexity axiom is insufficient for characterizing CHSH correlations. Both results build on new insights about reliable classical computation. The first exploits our formalization of an equivalence between amplification and reliable computation, while the second follows from an upper bound on the threshold for reliable computation with formulas of noisy XOR and AND gates.

cs.IT↗

arXiv2023

Max-Margin Works while Large Margin Fails: Generalization without Uniform Convergence

A major challenge in modern machine learning is theoretically understanding the generalization properties of overparameterized models. Many existing tools rely on uniform convergence (UC), a property that, when it holds, guarantees that the test loss will be close to the training loss, uniformly over a class of candidate models. Nagarajan and Kolter (2019) show that in certain simple linear and neural-network settings, any uniform convergence bound will be vacuous, leaving open the question of how to prove generalization in settings where UC fails. Our main contribution is proving novel generalization bounds in two such settings, one linear, and one non-linear. We study the linear classification setting of Nagarajan and Kolter, and a quadratic ground truth function learned via a two-layer neural network in the non-linear regime. We prove a new type of margin bound showing that above a certain signal-to-noise threshold, any near-max-margin classifier will achieve almost no test loss in these two settings. Our results show that near-max-margin is important: while any model that achieves at least a $(1 - ε)$-fraction of the max-margin generalizes well, a classifier achieving half of the max-margin may fail terribly. Building on the impossibility results of Nagarajan and Kolter, under slightly stronger assumptions, we show that one-sided UC bounds and classical margin bounds will fail on near-max-margin classifiers. Our analysis provides insight on why memorization can coexist with generalization: we show that in this challenging regime where generalization occurs but UC fails, near-max-margin classifiers simultaneously contain some generalizable components and some overfitting components that memorize the data. The presence of the overfitting components is enough to preclude UC, but the near-extremal margin guarantees that sufficient generalizable components are present.

cs.LG↗

arXiv2022

On the Download Rate of Homomorphic Secret Sharing

A homomorphic secret sharing (HSS) scheme is a secret sharing scheme that supports evaluating functions on shared secrets by means of a local mapping from input shares to output shares. We initiate the study of the download rate of HSS, namely, the achievable ratio between the length of the output shares and the output length when amortized over $\ell$ function evaluations. We obtain the following results. * In the case of linear information-theoretic HSS schemes for degree-$d$ multivariate polynomials, we characterize the optimal download rate in terms of the optimal minimal distance of a linear code with related parameters. We further show that for sufficiently large $\ell$ (polynomial in all problem parameters), the optimal rate can be realized using Shamir's scheme, even with secrets over $\mathbb{F}_2$. * We present a general rate-amplification technique for HSS that improves the download rate at the cost of requiring more shares. As a corollary, we get high-rate variants of computationally secure HSS schemes and efficient private information retrieval protocols from the literature. * We show that, in some cases, one can beat the best download rate of linear HSS by allowing nonlinear output reconstruction and $2^{-Ω(\ell)}$ error probability.

cs.IT↗

arXiv2022

Efficient Near-Optimal Codes for General Repeat Channels

Given a probability distribution $\mathcal{D}$ over the non-negative integers, a $\mathcal{D}$-repeat channel acts on an input symbol by repeating it a number of times distributed as $\mathcal{D}$. For example, the binary deletion channel ($\mathcal{D}=Bernoulli$) and the Poisson repeat channel ($\mathcal{D}=Poisson$) are special cases. We say a $\mathcal{D}$-repeat channel is square-integrable if $\mathcal{D}$ has finite first and second moments. In this paper, we construct explicit codes for all square-integrable $\mathcal{D}$-repeat channels with rate arbitrarily close to the capacity, that are encodable and decodable in linear and quasi-linear time, respectively. We also consider possible extensions to the repeat channel model, and illustrate how our construction can be extended to an even broader class of channels capturing insertions, deletions, and substitutions. Our work offers an alternative, simplified, and more general construction to the recent work of Rubinstein (arXiv:2111.00261), who attains similar results to ours in the cases of the deletion channel and the Poisson repeat channel. It also slightly improves the runtime and decoding failure probability of the polar codes constructions of Tal et al. (ISIT 2019) and of Pfister and Tal (arXiv:2102.02155) for the deletion channel and certain insertion/deletion/substitution channels. Our techniques follow closely the approaches of Guruswami and Li (IEEEToIT 2019) and Con and Shpilka (IEEEToIT 2020); what sets apart our work is that we show that a capacity-achieving code can be assumed to have an "approximate balance" in the frequency of zeros and ones of all sufficiently long substrings of all codewords. This allows us to attain near-capacity-achieving codes in a general setting. We consider this "approximate balance" result to be of independent interest, as it can be cast in much greater generality than repeat channels.

cs.IT↗

arXiv2022

Low-bandwidth recovery of linear functions of Reed-Solomon-encoded data

We study the problem of efficiently computing on encoded data. More specifically, we study the question of low-bandwidth computation of functions $F:\mathbb{F}^k \to \mathbb{F}$ of some data $x \in \mathbb{F}^k$, given access to an encoding $c \in \mathbb{F}^n$ of $x$ under an error correcting code. In our model -- relevant in distributed storage, distributed computation and secret sharing -- each symbol of $c$ is held by a different party, and we aim to minimize the total amount of information downloaded from each party in order to compute $F(x)$. Special cases of this problem have arisen in several domains, and we believe that it is fruitful to study this problem in generality. Our main result is a low-bandwidth scheme to compute linear functions for Reed-Solomon codes, even in the presence of erasures. More precisely, let $ε> 0$ and let $\mathcal{C}: \mathbb{F}^k \to \mathbb{F}^n$ be a full-length Reed-Solomon code of rate $1 - ε$ over a field $\mathbb{F}$ with constant characteristic. For any $γ\in [0, ε)$, our scheme can compute any linear function $F(x)$ given access to any $(1 - γ)$-fraction of the symbols of $\mathcal{C}(x)$, with download bandwidth $O(n/(ε- γ))$ bits. In contrast, the naive scheme that involves reconstructing the data $x$ and then computing $F(x)$ uses $Θ(n \log n)$ bits. Our scheme has applications in distributed storage, coded computation, and homomorphic secret sharing.

cs.IT↗

arXiv2021

LDPC Codes Achieve List Decoding Capacity

We show that Gallager's ensemble of Low-Density Parity Check (LDPC) codes achieves list-decoding capacity with high probability. These are the first graph-based codes shown to have this property. This result opens up a potential avenue towards truly linear-time list-decodable codes that achieve list-decoding capacity. Our result on list decoding follows from a much more general result: any $\textit{local}$ property satisfied with high probability by a random linear code is also satisfied with high probability by a random LDPC code from Gallager's distribution. Local properties are properties characterized by the exclusion of small sets of codewords, and include list-decodability, list-recoverability and average-radius list-decodability. In order to prove our results on LDPC codes, we establish sharp thresholds for when local properties are satisfied by a random linear code. More precisely, we show that for any local property $\mathcal{P}$, there is some $R^*$ so that random linear codes of rate slightly less than $R^*$ satisfy $\mathcal{P}$ with high probability, while random linear codes of rate slightly more than $R^*$, with high probability, do not. We also give a characterization of the threshold rate $R^*$.

cs.IT↗

arXiv2021

A Note on the Permuted Puzzles Toy Conjecture

In this note, we show that a "Toy Conjecture" made by (Boyle, Ishai, Pass, Wootters, 2017) is false, and propose a new one. Our attack does not falsify the full ("non-toy") conjecture in that work, and it is our hope that this note will help further the analysis of that conjecture. Independently, (Boyle, Holmgren, Ma, Weiss, 2021) have obtained similar results.

cs.CR↗

arXiv2021

Wedge-Lifted Codes

We define wedge-lifted codes, a variant of lifted codes, and we study their locality properties. We show that (taking the trace of) wedge-lifted codes yields binary codes with the $t$-disjoint repair property ($t$-DRGP). When $t = N^{1/2d}$, where $N$ is the block length of the code and $d \geq 2$ is any integer, our codes give improved trade-offs between redundancy and locality among binary codes.

cs.IT↗

arXiv2021

Approximate Gradient Coding with Optimal Decoding

In distributed optimization problems, a technique called gradient coding, which involves replicating data points, has been used to mitigate the effect of straggling machines. Recent work has studied approximate gradient coding, which concerns coding schemes where the replication factor of the data is too low to recover the full gradient exactly. Our work is motivated by the challenge of creating approximate gradient coding schemes that simultaneously work well in both the adversarial and stochastic models. To that end, we introduce novel approximate gradient codes based on expander graphs, in which each machine receives exactly two blocks of data points. We analyze the decoding error both in the random and adversarial straggler setting, when optimal decoding coefficients are used. We show that in the random setting, our schemes achieve an error to the gradient that decays exponentially in the replication factor. In the adversarial setting, the error is nearly a factor of two smaller than any existing code with similar performance in the random setting. We show convergence bounds both in the random and adversarial setting for gradient descent under standard assumptions using our codes. In the random setting, our convergence rate improves upon block-box bounds. In the adversarial setting, we show that gradient descent can converge down to a noise floor that scales linearly with the adversarial error to the gradient. We demonstrate empirically that our schemes achieve near-optimal error in the random setting and converge faster than algorithms which do not use the optimal decoding coefficients.

stat.ML↗

arXiv2021

Improved batch code lower bounds

Batch codes are a useful notion of locality for error correcting codes, originally introduced in the context of distributed storage and cryptography. Many constructions of batch codes have been given, but few lower bound (limitation) results are known, leaving gaps between the best known constructions and best known lower bounds. Towards determining the optimal redundancy of batch codes, we prove a new lower bound on the redundancy of batch codes. Specifically, we study (primitive, multiset) linear batch codes that systematically encode $n$ information symbols into $N$ codeword symbols, with the requirement that any multiset of $k$ symbol requests can be obtained in disjoint ways. We show that such batch codes need $Ω(\sqrt{Nk})$ symbols of redundancy, improving on the previous best lower bounds of $Ω(\sqrt{N}+k)$ at all $k=n^\varepsilon$ with $\varepsilon\in(0,1)$. Our proof follows from analyzing the dimension of the order-$O(k)$ tensor of the batch code's dual code.

cs.IT↗

arXiv2021

Asynchronous Distributed Optimization with Stochastic Delays

We study asynchronous finite sum minimization in a distributed-data setting with a central parameter server. While asynchrony is well understood in parallel settings where the data is accessible by all machines -- e.g., modifications of variance-reduced gradient algorithms like SAGA work well -- little is known for the distributed-data setting. We develop an algorithm ADSAGA based on SAGA for the distributed-data setting, in which the data is partitioned between many machines. We show that with $m$ machines, under a natural stochastic delay model with an mean delay of $m$, ADSAGA converges in $\tilde{O}\left(\left(n + \sqrt{m}κ\right)\log(1/ε)\right)$ iterations, where $n$ is the number of component functions, and $κ$ is a condition number. This complexity sits squarely between the complexity $\tilde{O}\left(\left(n + κ\right)\log(1/ε)\right)$ of SAGA \textit{without delays} and the complexity $\tilde{O}\left(\left(n + mκ\right)\log(1/ε)\right)$ of parallel asynchronous algorithms where the delays are \textit{arbitrary} (but bounded by $O(m)$), and the data is accessible by all. Existing asynchronous algorithms with distributed-data setting and arbitrary delays have only been shown to converge in $\tilde{O}(n^2κ\log(1/ε))$ iterations. We empirically compare on least-squares problems the iteration complexity and wallclock performance of ADSAGA to existing parallel and distributed algorithms, including synchronous minibatch algorithms. Our results demonstrate the wallclock advantage of variance-reduced asynchronous approaches over SGD or synchronous approaches.

cs.LG↗

arXiv2021

On Greedy Approaches to Hierarchical Aggregation

We analyze greedy algorithms for the Hierarchical Aggregation (HAG) problem, a strategy introduced in [Jia et al., KDD 2020] for speeding up learning on Graph Neural Networks (GNNs). The idea of HAG is to identify and remove redundancies in computations performed when training GNNs. The associated optimization problem is to identify and remove the most redundancies. Previous work introduced a greedy approach for the HAG problem and claimed a 1-1/e approximation factor. We show by example that this is not correct, and one cannot hope for better than a 1/2 approximation factor. We prove that this greedy algorithm does satisfy some (weaker) approximation guarantee, by showing a new connection between the HAG problem and maximum matching problems in hypergraphs. We also introduce a second greedy algorithm which can out-perform the first one, and we show how to implement it efficiently in some parameter regimes. Finally, we introduce some greedy heuristics that are much faster than the above greedy algorithms, and we demonstrate that they perform well on real-world graphs.

cs.DS↗

arXiv2021

On Coding for an Abstracted Nanopore Channel for DNA Storage

In the emerging field of DNA storage, data is encoded as DNA sequences and stored. The data is read out again by sequencing the stored DNA. Nanopore sequencing is a new sequencing technology that has many advantages over other methods; in particular, it is cheap, portable, and can support longer reads. While several practical coding schemes have been developed for DNA storage with nanopore sequencing, the theory is not well understood. Towards that end, we study a highly abstracted (deterministic) version of the nanopore sequencer, which highlights key features that make its analysis difficult. We develop methods and theory to understand the capacity of our abstracted model, and we propose efficient coding schemes and algorithms.

cs.IT↗

arXiv2020

Improved list-decodability of random linear binary codes

There has been a great deal of work establishing that random linear codes are as list-decodable as uniformly random codes, in the sense that a random linear binary code of rate $1 - H(p) - ε$ is $(p,O(1/ε))$-list-decodable with high probability. In this work, we show that such codes are $(p, H(p)/ε+ 2)$-list-decodable with high probability, for any $p \in (0, 1/2)$ and $ε> 0$. In addition to improving the constant in known list-size bounds, our argument, which is quite simple, works simultaneously for all values of $p$, while previous works obtaining $L = O(1/ε)$ patched together different arguments to cover different parameter regimes. Our approach is to strengthen an existential argument of (Guruswami, Håstad, Sudan and Zuckerman, IEEE Trans. IT, 2002) to hold with high probability. To complement our upper bound for random linear codes, we also improve an argument of (Guruswami, Narayanan, IEEE Trans. IT, 2014) to obtain an essentially tight lower bound of $1/ε$ on the list size of uniformly random codes; this implies that random linear codes are in fact more list-decodable than uniformly random codes, in the sense that the list sizes are strictly smaller. To demonstrate the applicability of these techniques, we use them to (a) obtain more information about the distribution of list sizes of random linear codes and (b) to prove a similar result for random linear rank-metric codes.

cs.IT↗

arXiv2020

Superbridge and Bridge Indices for Knots

We improve the upper bound on the superbridge index $sb[K]$ of a knot type $[K]$ in terms of the bridge index $b[K]$ from $sb[K] \leq 5b -3$ to $sb[K]\leq 3b[k] - 1$.

math.GT↗

arXiv2020

Lifted multiplicity codes and the disjoint repair group property

Lifted Reed Solomon Codes (Guo, Kopparty, Sudan 2013) were introduced in the context of locally correctable and testable codes. They are multivariate polynomials whose restriction to any line is a codeword of a Reed-Solomon code. We consider a generalization of their construction, which we call lifted multiplicity codes. These are multivariate polynomial codes whose restriction to any line is a codeword of a multiplicity code (Kopparty, Saraf, Yekhanin 2014). We show that lifted multiplicity codes have a better trade-off between redundancy and a notion of locality called the $t$-disjoint-repair-group property than previously known constructions. More precisely, we show that lifted multiplicity codes with length $N$ and redundancy $O(t^{0.585} \sqrt{N})$ have the property that any symbol of a codeword can be reconstructed in $t$ different ways, each using a disjoint subset of the other coordinates. This gives the best known trade-off for this problem for any super-constant $t < \sqrt{N}$. We also give an alternative analysis of lifted Reed Solomon codes using dual codes, which may be of independent interest.

cs.IT↗