arXiv subjects

Nisheeth K. Vishnoi

Publications and source records attributed to Nisheeth K. Vishnoi.

At least 55 records · Page 3Linked to original sources

arXiv2020

Data preprocessing to mitigate bias: A maximum entropy based approach

Data containing human or social attributes may over- or under-represent groups with respect to salient social attributes such as gender or race, which can lead to biases in downstream applications. This paper presents an algorithmic framework that can be used as a data preprocessing method towards mitigating such bias. Unlike prior work, it can efficiently learn distributions over large domains, controllably adjust the representation rates of protected groups and achieve target fairness metrics such as statistical parity, yet remains close to the empirical distribution induced by the given dataset. Our approach leverages the principle of maximum entropy - amongst all distributions satisfying a given set of constraints, we should choose the one closest in KL-divergence to a given prior. While maximum entropy distributions can succinctly encode distributions over large domains, they can be difficult to compute. Our main contribution is an instantiation of this framework for our set of constraints and priors, which encode our bias mitigation goals, and that runs in time polynomial in the dimension of the data. Empirically, we observe that samples from the learned distribution have desired representation rates and statistical rates, and when used for training a classifier incurs only a slight loss in accuracy while maintaining fairness properties.

cs.LG↗

arXiv2020

Coresets for Clustering in Euclidean Spaces: Importance Sampling is Nearly Optimal

Given a collection of $n$ points in $\mathbb{R}^d$, the goal of the $(k,z)$-clustering problem is to find a subset of $k$ "centers" that minimizes the sum of the $z$-th powers of the Euclidean distance of each point to the closest center. Special cases of the $(k,z)$-clustering problem include the $k$-median and $k$-means problems. Our main result is a unified two-stage importance sampling framework that constructs an $\varepsilon$-coreset for the $(k,z)$-clustering problem. Compared to the results for $(k,z)$-clustering in [Feldman and Langberg, STOC 2011], our framework saves a $\varepsilon^2 d$ factor in the coreset size. Compared to the results for $(k,z)$-clustering in [Sohler and Woodruff, FOCS 2018], our framework saves a $\operatorname{poly}(k)$ factor in the coreset size and avoids the $\exp(k/\varepsilon)$ term in the construction time. Specifically, our coreset for $k$-median ($z=1$) has size $\tilde{O}(\varepsilon^{-4} k)$ which, when compared to the result in [Sohler and Woodruff, STOC 2018], saves a $k$ factor in the coreset size. Our algorithmic results rely on a new dimensionality reduction technique that connects two well-known shape fitting problems: subspace approximation and clustering, and may be of independent interest. We also provide a size lower bound of $Ω\left(k\cdot \min \left\{2^{z/20},d \right\}\right)$ for a $0.01$-coreset for $(k,z)$-clustering, which has a linear dependence of size on $k$ and an exponential dependence on $z$ that matches our algorithmic results.

cs.CG↗

arXiv2020

On the computability of continuous maximum entropy distributions with applications

We initiate a study of the following problem: Given a continuous domain $Ω$ along with its convex hull $\mathcal{K}$, a point $A \in \mathcal{K}$ and a prior measure $μ$ on $Ω$, find the probability density over $Ω$ whose marginal is $A$ and that minimizes the KL-divergence to $μ$. This framework gives rise to several extremal distributions that arise in mathematics, quantum mechanics, statistics, and theoretical computer science. Our technical contributions include a polynomial bound on the norm of the optimizer of the dual problem that holds in a very general setting and relies on a "balance" property of the measure $μ$ on $Ω$, and exact algorithms for evaluating the dual and its gradient for several interesting settings of $Ω$ and $μ$. Together, along with the ellipsoid method, these results imply polynomial-time algorithms to compute such KL-divergence minimizing distributions in several cases. Applications of our results include: 1) an optimization characterization of the Goemans-Williamson measure that is used to round a positive semidefinite matrix to a vector, 2) the computability of the entropic barrier for polytopes studied by Bubeck and Eldan, and 3) a polynomial-time algorithm to compute the barycentric quantum entropy of a density matrix that was proposed as an alternative to von Neumann entropy in the 1970s: this corresponds to the case when $Ω$ is the set of rank one projections matrices and $μ$ corresponds to the Haar measure on the unit sphere. Our techniques generalize to the setting of Hermitian rank $k$ projections using the Harish-Chandra-Itzykson-Zuber formula, and are applicable even beyond, to adjoint orbits of compact Lie groups.

cs.DS↗

arXiv2020

Classification with Fairness Constraints: A Meta-Algorithm with Provable Guarantees

Developing classification algorithms that are fair with respect to sensitive attributes of the data has become an important problem due to the growing deployment of classification algorithms in various social contexts. Several recent works have focused on fairness with respect to a specific metric, modeled the corresponding fair classification problem as a constrained optimization problem, and developed tailored algorithms to solve them. Despite this, there still remain important metrics for which we do not have fair classifiers and many of the aforementioned algorithms do not come with theoretical guarantees; perhaps because the resulting optimization problem is non-convex. The main contribution of this paper is a new meta-algorithm for classification that takes as input a large class of fairness constraints, with respect to multiple non-disjoint sensitive attributes, and which comes with provable guarantees. This is achieved by first developing a meta-algorithm for a large family of classification problems with convex constraints, and then showing that classification problems with general types of fairness constraints can be reduced to those in this family. We present empirical results that show that our algorithm can achieve near-perfect fairness with respect to various fairness metrics, and that the loss in accuracy due to the imposed fairness constraints is often small. Overall, this work unifies several prior works on fair classification, presents a practical algorithm with theoretical guarantees, and can handle fairness metrics that were previously not possible.

cs.LG↗

arXiv2020

Interventions for Ranking in the Presence of Implicit Bias

Implicit bias is the unconscious attribution of particular qualities (or lack thereof) to a member from a particular social group (e.g., defined by gender or race). Studies on implicit bias have shown that these unconscious stereotypes can have adverse outcomes in various social contexts, such as job screening, teaching, or policing. Recently, (Kleinberg and Raghavan, 2018) considered a mathematical model for implicit bias and showed the effectiveness of the Rooney Rule as a constraint to improve the utility of the outcome for certain cases of the subset selection problem. Here we study the problem of designing interventions for the generalization of subset selection -- ranking -- that requires to output an ordered set and is a central primitive in various social and computational contexts. We present a family of simple and interpretable constraints and show that they can optimally mitigate implicit bias for a generalization of the model studied in (Kleinberg and Raghavan, 2018). Subsequently, we prove that under natural distributional assumptions on the utilities of items, simple, Rooney Rule-like, constraints can also surprisingly recover almost all the utility lost due to implicit biases. Finally, we augment our theoretical results with empirical findings on real-world distributions from the IIT-JEE (2009) dataset and the Semantic Scholar Research corpus.

cs.CY↗

arXiv2019

Coresets for Clustering with Fairness Constraints

In a recent work, [19] studied the following "fair" variants of classical clustering problems such as $k$-means and $k$-median: given a set of $n$ data points in $\mathbb{R}^d$ and a binary type associated to each data point, the goal is to cluster the points while ensuring that the proportion of each type in each cluster is roughly the same as its underlying proportion. Subsequent work has focused on either extending this setting to when each data point has multiple, non-disjoint sensitive types such as race and gender [6], or to address the problem that the clustering algorithms in the above work do not scale well. The main contribution of this paper is an approach to clustering with fairness constraints that involve multiple, non-disjoint types, that is also scalable. Our approach is based on novel constructions of coresets: for the $k$-median objective, we construct an $\varepsilon$-coreset of size $O(Γk^2 \varepsilon^{-d})$ where $Γ$ is the number of distinct collections of groups that a point may belong to, and for the $k$-means objective, we show how to construct an $\varepsilon$-coreset of size $O(Γk^3\varepsilon^{-d-1})$. The former result is the first known coreset construction for the fair clustering problem with the $k$-median objective, and the latter result removes the dependence on the size of the full dataset as in [39] and generalizes it to multiple, non-disjoint types. Plugging our coresets into existing algorithms for fair clustering such as [5] results in the fastest algorithms for several cases. Empirically, we assess our approach over the \textbf{Adult}, \textbf{Bank}, \textbf{Diabetes} and \textbf{Athlete} dataset, and show that the coreset sizes are much smaller than the full dataset. We also achieve a speed-up to recent fair clustering algorithms [5,6] by incorporating our coreset construction.

cs.DS↗

arXiv2019

Online Sampling from Log-Concave Distributions

Given a sequence of convex functions $f_0, f_1, \ldots, f_T$, we study the problem of sampling from the Gibbs distribution $π_t \propto e^{-\sum_{k=0}^tf_k}$ for each epoch $t$ in an online manner. Interest in this problem derives from applications in machine learning, Bayesian statistics, and optimization where, rather than obtaining all the observations at once, one constantly acquires new data, and must continuously update the distribution. Our main result is an algorithm that generates roughly independent samples from $π_t$ for every epoch $t$ and, under mild assumptions, makes $\mathrm{polylog}(T)$ gradient evaluations per epoch. All previous results imply a bound on the number of gradient or function evaluations which is at least linear in $T$. Motivated by real-world applications, we assume that functions are smooth, their associated distributions have a bounded second moment, and their minimizer drifts in a bounded manner, but do not assume they are strongly convex. In particular, our assumptions hold for online Bayesian logistic regression, when the data satisfy natural regularity properties, giving a sampling algorithm with updates that are poly-logarithmic in $T$. In simulations, our algorithm achieves accuracy comparable to an algorithm specialized to logistic regression. Key to our algorithm is a novel stochastic gradient Langevin dynamics Markov chain with a carefully designed variance reduction step and constant batch size. Technically, lack of strong convexity is a significant barrier to analysis and, here, our main contribution is a martingale exit time argument that shows our Markov chain remains in a ball of radius roughly poly-logarithmic in $T$ for enough time to reach within $\varepsilon$ of $π_t$.

cs.LG↗

arXiv2019

Faster polytope rounding, sampling, and volume computation via a sublinear "Ball Walk"

We study the problem of "isotropically rounding" a polytope $K\subset\mathbb{R}^n$, that is, computing a linear transformation which makes the uniform distribution on the polytope have roughly identity covariance matrix. We assume $K$ is defined by $m$ linear inequalities, with guarantee that $rB\subset K\subset RB$, where $B$ is the unit ball. We introduce a new variant of the ball walk Markov chain and show that, roughly, the expected number of arithmetic operations per-step of this Markov chain is $O(m)$ that is sublinear in the input size $mn$--the per-step time of all prior Markov chains. Subsequently, we give a rounding algorithm that succeeds with probability $1-\varepsilon$ in $\tilde{O}(mn^{4.5}\mbox{polylog}(\frac{1}{\varepsilon},\frac{R}{r}))$ arithmetic operations. This gives a factor of $\sqrt{n}$ improvement on the previous bound of $\tilde{O}(mn^5\mbox{polylog}(\frac{1}{\varepsilon},\frac{R}{r}))$ for rounding, which uses the hit-and-run algorithm. Since the rounding preprocessing step is in many cases the bottleneck in improving sampling or volume computation, our results imply these tasks can also be achieved in roughly $\tilde{O}(mn^{4.5}\mbox{polylog}(\frac{1}{\varepsilon},\frac{R}{r})+mn^4δ^{-2})$ operations for computing the volume of $K$ up to a factor $1+δ$ and $\tilde{O}(mn^{4.5}\mbox{polylog}(\frac{1}{\varepsilon},\frac{R}{r})))$ for uniformly sampling on $K$ with TV error $\varepsilon$. This improves on the previous bounds of $\tilde{O}(mn^5\mbox{polylog}(\frac{1}{\varepsilon},\frac{R}{r})+mn^4δ^{-2})$ for volume computation when roughly $m\geq n^{2.5}$, and $\tilde{O}(mn^5\mbox{polylog}(\frac{1}{\varepsilon},\frac{R}{r}))$ for sampling when roughly $m\geq n^{1.5}$. We achieve this improvement by a novel method of computing polytope membership, where one avoids checking inequalities estimated to have a very low probability of being violated.

cs.DS↗

arXiv2019

Maximum Entropy Distributions: Bit Complexity and Stability

Maximum entropy distributions with discrete support in $m$ dimensions arise in machine learning, statistics, information theory, and theoretical computer science. While structural and computational properties of max-entropy distributions have been extensively studied, basic questions such as: Do max-entropy distributions over a large support (e.g., $2^m$) with a specified marginal vector have succinct descriptions (polynomial-size in the input description)? and: Are entropy maximizing distributions "stable" under the perturbation of the marginal vector? have resisted a rigorous resolution. Here we show that these questions are related and resolve both of them. Our main result shows a ${\rm poly}(m, \log 1/\varepsilon)$ bound on the bit complexity of $\varepsilon$-optimal dual solutions to the maximum entropy convex program -- for very general support sets and with no restriction on the marginal vector. Applications of this result include polynomial time algorithms to compute max-entropy distributions over several new and old polytopes for any marginal vector in a unified manner, a polynomial time algorithm to compute the Brascamp-Lieb constant in the rank-1 case. The proof of this result allows us to show that changing the marginal vector by $δ$ changes the max-entropy distribution in the total variation distance roughly by a factor of ${\rm poly}(m, \log 1/δ)\sqrtδ$ -- even when the size of the support set is exponential. Together, our results put max-entropy distributions on a mathematically sound footing -- these distributions are robust and computationally feasible models for data.

cs.DS↗

arXiv2019

Toward Controlling Discrimination in Online Ad Auctions

Online advertising platforms are thriving due to the customizable audiences they offer advertisers. However, recent studies show that advertisements can be discriminatory with respect to the gender or race of the audience that sees the ad, and may inadvertently cross ethical and/or legal boundaries. To prevent this, we propose a constrained ad auction framework that maximizes the platform's revenue conditioned on ensuring that the audience seeing an advertiser's ad is distributed appropriately across sensitive types such as gender or race. Building upon Myerson's classic work, we first present an optimal auction mechanism for a large class of fairness constraints. Finding the parameters of this optimal auction, however, turns out to be a non-convex problem. We show that this non-convex problem can be reformulated as a more structured non-convex problem with no saddle points or local-maxima; this allows us to develop a gradient-descent-based algorithm to solve it. Our empirical results on the A1 Yahoo! dataset demonstrate that our algorithm can obtain uniform coverage across different user types for each advertiser at a minor loss to the revenue of the platform, and a small change to the size of the audience each advertiser reaches.

cs.GT↗

arXiv2019

Nonconvex sampling with the Metropolis-adjusted Langevin algorithm

The Langevin Markov chain algorithms are widely deployed methods to sample from distributions in challenging high-dimensional and non-convex statistics and machine learning applications. Despite this, current bounds for the Langevin algorithms are slower than those of competing algorithms in many important situations, for instance when sampling from weakly log-concave distributions, or when sampling or optimizing non-convex log-densities. In this paper, we obtain improved bounds in many of these situations, showing that the Metropolis-adjusted Langevin algorithm (MALA) is faster than the best bounds for its competitor algorithms when the target distribution satisfies weak third- and fourth- order regularity properties associated with the input data. In many settings, our regularity conditions are weaker than the usual Euclidean operator norm regularity properties, allowing us to show faster bounds for a much larger class of distributions than would be possible with the usual Euclidean operator norm approach, including in statistics and machine learning applications where the data satisfy a certain incoherence condition. In particular, we show that using our regularity conditions one can obtain faster bounds for applications which include sampling problems in Bayesian logistic regression with weakly convex priors, and the nonconvex optimization problem of learning linear classifiers with zero-one loss functions. Our main technical contribution in this paper is our analysis of the Metropolis acceptance probability of MALA in terms of its "energy-conservation error," and our bound for this error in terms of third- and fourth- order regularity conditions. Our combination of this higher-order analysis of the energy conservation error with the conductance method is key to obtaining bounds which have a sub-linear dependence on the dimension $d$ in the non-strongly logconcave setting.

cs.DS↗

arXiv2018

On the Number of Circuits in Regular Matroids (with Connections to Lattices and Codes)

We show that for any regular matroid on $m$ elements and any $α\geq 1$, the number of $α$-minimum circuits, or circuits whose size is at most an $α$-multiple of the minimum size of a circuit in the matroid is bounded by $m^{O(α^2)}$. This generalizes a result of Karger for the number of $α$-minimum cuts in a graph. As a consequence, we obtain similar bounds on the number of $α$-shortest vectors in "totally unimodular" lattices and on the number of $α$-minimum weight codewords in "regular" codes.

cs.DS↗

arXiv2018

Dynamic Sampling from Graphical Models

In this paper, we study the problem of sampling from a graphical model when the model itself is changing dynamically with time. This problem derives its interest from a variety of inference, learning, and sampling settings in machine learning, computer vision, statistical physics, and theoretical computer science. While the problem of sampling from a static graphical model has received considerable attention, theoretical works for its dynamic variants have been largely lacking. The main contribution of this paper is an algorithm that can sample dynamically from a broad class of graphical models over discrete random variables. Our algorithm is parallel and Las Vegas: it knows when to stop and it outputs samples from the exact distribution. We also provide sufficient conditions under which this algorithm runs in time proportional to the size of the update, on general graphical models as well as well-studied specific spin systems. In particular we obtain, for the Ising model (ferromagnetic or anti-ferromagnetic) and for the hardcore model the first dynamic sampling algorithms that can handle both edge and vertex updates (addition, deletion, change of functions), both efficient within regimes that are close to the respective uniqueness regimes, beyond which, even for the static and approximate sampling, no local algorithms were known or the problem itself is intractable. Our dynamic sampling algorithm relies on a local resampling algorithm and a new "equilibrium" property that is shown to be satisfied by our algorithm at each step, and enables us to prove its correctness. This equilibrium property is robust enough to guarantee the correctness of our algorithm, helps us improve bounds on fast convergence on specific models, and should be of independent interest.

cs.DS↗

arXiv2018

Dimensionally Tight Bounds for Second-Order Hamiltonian Monte Carlo

Hamiltonian Monte Carlo (HMC) is a widely deployed method to sample from high-dimensional distributions in Statistics and Machine learning. HMC is known to run very efficiently in practice and its popular second-order "leapfrog" implementation has long been conjectured to run in $d^{1/4}$ gradient evaluations. Here we show that this conjecture is true when sampling from strongly log-concave target distributions that satisfy a weak third-order regularity property associated with the input data. Our regularity condition is weaker than the Lipschitz Hessian property and allows us to show faster convergence bounds for a much larger class of distributions than would be possible with the usual Lipschitz Hessian constant alone. Important distributions that satisfy our regularity condition include posterior distributions used in Bayesian logistic regression for which the data satisfies an "incoherence" property. Our result compares favorably with the best available bounds for the class of strongly log-concave distributions, which grow like $d^{{1}/{2}}$ gradient evaluations with the dimension. Moreover, our simulations on synthetic data suggest that, when our regularity condition is satisfied, leapfrog HMC performs better than its competitors -- both in terms of accuracy and in terms of the number of gradient evaluations it requires.

cs.DS↗

arXiv2018

Ranking with Fairness Constraints

Ranking algorithms are deployed widely to order a set of items in applications such as search engines, news feeds, and recommendation systems. Recent studies, however, have shown that, left unchecked, the output of ranking algorithms can result in decreased diversity in the type of content presented, promote stereotypes, and polarize opinions. In order to address such issues, we study the following variant of the traditional ranking problem when, in addition, there are fairness or diversity constraints. Given a collection of items along with 1) the value of placing an item in a particular position in the ranking, 2) the collection of sensitive attributes (such as gender, race, political opinion) of each item and 3) a collection of constraints that, for each k, bound the number of items with each attribute that are allowed to appear in the top k positions of the ranking, the goal is to output a ranking that maximizes the value with respect to the original rank quality metric while respecting the constraints. This problem encapsulates various well-studied problems related to bipartite and hypergraph matching as special cases and turns out to be hard to approximate even with simple constraints. Our main technical contributions are fast exact and approximation algorithms along with complementary hardness results that, together, come close to settling the approximability of this constrained ranking maximization problem. Unlike prior work on the constrained matching problems, our algorithm runs in linear time, even when the number of constraints is large, its approximation ratio does not depend on the number of constraints, and it produces solutions with small constraint violations. Our results rely on insights about the constrained matching problem when the objective satisfies properties that appear in common ranking metrics such as Discounted Cumulative Gain, Spearman's rho or Bradley-Terry.

cs.DS↗

arXiv2018

Subdeterminant Maximization via Nonconvex Relaxations and Anti-concentration

Several fundamental problems that arise in optimization and computer science can be cast as follows: Given vectors $v_1,\ldots,v_m \in \mathbb{R}^d$ and a constraint family ${\cal B}\subseteq 2^{[m]}$, find a set $S \in \cal{B}$ that maximizes the squared volume of the simplex spanned by the vectors in $S$. A motivating example is the data-summarization problem in machine learning where one is given a collection of vectors that represent data such as documents or images. The volume of a set of vectors is used as a measure of their diversity, and partition or matroid constraints over $[m]$ are imposed in order to ensure resource or fairness constraints. Recently, Nikolov and Singh presented a convex program and showed how it can be used to estimate the value of the most diverse set when ${\cal B}$ corresponds to a partition matroid. This result was recently extended to regular matroids in works of Straszak and Vishnoi, and Anari and Oveis Gharan. The question of whether these estimation algorithms can be converted into the more useful approximation algorithms -- that also output a set -- remained open. The main contribution of this paper is to give the first approximation algorithms for both partition and regular matroids. We present novel formulations for the subdeterminant maximization problem for these matroids; this reduces them to the problem of finding a point that maximizes the absolute value of a nonconvex function over a Cartesian product of probability simplices. The technical core of our results is a new anti-concentration inequality for dependent random variables that allows us to relate the optimal value of these nonconvex functions to their value at a random point. Unlike prior work on the constrained subdeterminant maximization problem, our proofs do not rely on real-stability or convexity and could be of independent interest both in algorithms and complexity.

cs.DS↗

arXiv2018

Balanced News Using Constrained Bandit-based Personalization

We present a prototype for a news search engine that presents balanced viewpoints across liberal and conservative articles with the goal of de-polarizing content and allowing users to escape their filter bubble. The balancing is done according to flexible user-defined constraints, and leverages recent advances in constrained bandit optimization. We showcase our balanced news feed by displaying it side-by-side with the news feed produced by a traditional (polarized) feed.

cs.CY↗

arXiv2018

Multiwinner Voting with Fairness Constraints

Multiwinner voting rules are used to select a small representative subset of candidates or items from a larger set given the preferences of voters. However, if candidates have sensitive attributes such as gender or ethnicity (when selecting a committee), or specified types such as political leaning (when selecting a subset of news items), an algorithm that chooses a subset by optimizing a multiwinner voting rule may be unbalanced in its selection -- it may under or over represent a particular gender or political orientation in the examples above. We introduce an algorithmic framework for multiwinner voting problems when there is an additional requirement that the selected subset should be "fair" with respect to a given set of attributes. Our framework provides the flexibility to (1) specify fairness with respect to multiple, non-disjoint attributes (e.g., ethnicity and gender) and (2) specify a score function. We study the computational complexity of this constrained multiwinner voting problem for monotone and submodular score functions and present several approximation algorithms and matching hardness of approximation results for various attribute group structure and types of score functions. We also present simulations that suggest that adding fairness constraints may not affect the scores significantly when compared to the unconstrained case.

cs.CY↗