arXiv subjects

Sergei Vassilvitskii

Publications and source records attributed to Sergei Vassilvitskii.

51 records · Page 3Linked to original sources

arXiv2018

Fair Clustering Through Fairlets

We study the question of fair clustering under the {\em disparate impact} doctrine, where each protected class must have approximately equal representation in every cluster. We formulate the fair clustering problem under both the $k$-center and the $k$-median objectives, and show that even with two protected classes the problem is challenging, as the optimum solution can violate common conventions---for instance a point may no longer be assigned to its nearest cluster center! En route we introduce the concept of fairlets, which are minimal sets that satisfy fair representation while approximately preserving the clustering objective. We show that any fair clustering problem can be decomposed into first finding good fairlets, and then using existing machinery for traditional clustering algorithms. While finding good fairlets can be NP-hard, we proceed to obtain efficient approximation algorithms based on minimum cost flow. We empirically quantify the value of fair clustering on real-world datasets with sensitive attributes.

cs.LG↗

arXiv2017

Revenue Optimization with Approximate Bid Predictions

In the context of advertising auctions, finding good reserve prices is a notoriously challenging learning problem. This is due to the heterogeneity of ad opportunity types and the non-convexity of the objective function. In this work, we show how to reduce reserve price optimization to the standard setting of prediction under squared loss, a well understood problem in the learning community. We further bound the gap between the expected bid and revenue in terms of the average loss of the predictor. This is the first result that formally relates the revenue gained to the quality of a standard machine learned model.

cs.LG↗

arXiv2017

Statistical Cost Sharing

We study the cost sharing problem for cooperative games in situations where the cost function $C$ is not available via oracle queries, but must instead be derived from data, represented as tuples $(S, C(S))$, for different subsets $S$ of players. We formalize this approach, which we call statistical cost sharing, and consider the computation of the core and the Shapley value, when the tuples are drawn from some distribution $\mathcal{D}$. Previous work by Balcan et al. in this setting showed how to compute cost shares that satisfy the core property with high probability for limited classes of functions. We expand on their work and give an algorithm that computes such cost shares for any function with a non-empty core. We complement these results by proving an inapproximability lower bound for a weaker relaxation. We then turn our attention to the Shapley value. We first show that when cost functions come from the family of submodular functions with bounded curvature, $κ$, the Shapley value can be approximated from samples up to a $\sqrt{1 - κ}$ factor, and that the bound is tight. We then define statistical analogues of the Shapley axioms, and derive a notion of statistical Shapley value. We show that these can always be approximated arbitrarily well for general functions over any distribution $\mathcal{D}$.

cs.GT↗

arXiv2016

Submodular Optimization over Sliding Windows

Maximizing submodular functions under cardinality constraints lies at the core of numerous data mining and machine learning applications, including data diversification, data summarization, and coverage problems. In this work, we study this question in the context of data streams, where elements arrive one at a time, and we want to design low-memory and fast update-time algorithms that maintain a good solution. Specifically, we focus on the sliding window model, where we are asked to maintain a solution that considers only the last $W$ items. In this context, we provide the first non-trivial algorithm that maintains a provable approximation of the optimum using space sublinear in the size of the window. In particular we give a $\frac{1}{3} - ε$ approximation algorithm that uses space polylogarithmic in the spread of the values of the elements, $Φ$, and linear in the solution size $k$ for any constant $ε> 0$ . At the same time, processing each element only requires a polylogarithmic number of evaluations of the function itself. When a better approximation is desired, we show a different algorithm that, at the cost of using more memory, provides a $\frac{1}{2} - ε$ approximation and allows a tunable trade-off between average update time and space. This algorithm matches the best known approximation guarantees for submodular optimization in insertion-only streams, a less general formulation of the problem. We demonstrate the efficacy of the algorithms on a number of real world datasets, showing that their practical performance far exceeds the theoretical bounds. The algorithms preserve high quality solutions in streams with millions of items, while storing a negligible fraction of them.

cs.DS↗

arXiv2016

A Field Guide to Personalized Reserve Prices

We study the question of setting and testing reserve prices in single item auctions when the bidders are not identical. At a high level, there are two generalizations of the standard second price auction: in the lazy version we first determine the winner, and then apply reserve prices; in the eager version we first discard the bidders not meeting their reserves, and then determine the winner among the rest. We show that the two versions have dramatically different properties: lazy reserves are easy to optimize, and A/B test in production, whereas eager reserves always lead to higher welfare, but their optimization is NP-complete, and naive A/B testing will lead to incorrect conclusions. Despite their different characteristics, we show that the overall revenue for the two scenarios is always within a factor of 2 of each other, even in the presence of correlated bids. Moreover, we prove that the eager auction dominates the lazy auction on revenue whenever the bidders are independent or symmetric. We complement our theoretical results with simulations on real world data that show that even suboptimally set eager reserve prices are preferred from a revenue standpoint.

cs.GT↗

arXiv2015

Sketching, Embedding, and Dimensionality Reduction for Information Spaces

Information distances like the Hellinger distance and the Jensen-Shannon divergence have deep roots in information theory and machine learning. They are used extensively in data analysis especially when the objects being compared are high dimensional empirical probability distributions built from data. However, we lack common tools needed to actually use information distances in applications efficiently and at scale with any kind of provable guarantees. We can't sketch these distances easily, or embed them in better behaved spaces, or even reduce the dimensionality of the space while maintaining the probability structure of the data. In this paper, we build these tools for information distances---both for the Hellinger distance and Jensen--Shannon divergence, as well as related measures, like the $χ^2$ divergence. We first show that they can be sketched efficiently (i.e. up to multiplicative error in sublinear space) in the aggregate streaming model. This result is exponentially stronger than known upper bounds for sketching these distances in the strict turnstile streaming model. Second, we show a finite dimensionality embedding result for the Jensen-Shannon and $χ^2$ divergences that preserves pair wise distances. Finally we prove a dimensionality reduction result for the Hellinger, Jensen--Shannon, and $χ^2$ divergences that preserves the information geometry of the distributions (specifically, by retaining the simplex structure of the space). While our second result above already implies that these divergences can be explicitly embedded in Euclidean space, retaining the simplex structure is important because it allows us to continue doing inference in the reduced space. In essence, we preserve not just the distance structure but the underlying geometry of the space.

cs.DS↗

arXiv2014

Value of Targeting

We undertake a formal study of the value of targeting data to an advertiser. As expected, this value is increasing in the utility difference between realizations of the targeting data and the accuracy of the data, and depends on the distribution of competing bids. However, this value may vary non-monotonically with an advertiser's budget. Similarly, modeling the values as either private or correlated, or allowing other advertisers to also make use of the data, leads to unpredictable changes in the value of data. We address questions related to multiple data sources, show that utility of additional data may be non-monotonic, and provide tradeoffs between the quality and the price of data sources. In a game-theoretic setting, we show that advertisers may be worse off than if the data had not been available at all. We also ask whether a publisher can infer the value an advertiser would place on targeting data from the advertiser's bidding behavior and illustrate that this is impossible.

cs.GT↗

arXiv2012

Scalable K-Means++

Over half a century old and showing no signs of aging, k-means remains one of the most popular data processing algorithms. As is well-known, a proper initialization of k-means is crucial for obtaining a good final solution. The recently proposed k-means++ initialization algorithm achieves this, obtaining an initial set of centers that is provably close to the optimum solution. A major downside of the k-means++ is its inherent sequential nature, which limits its applicability to massive data: one must make k passes over the data to find a good initial set of centers. In this work we show how to drastically reduce the number of passes needed to obtain, in parallel, a good initialization. This is unlike prevailing efforts on parallelizing k-means that have mostly focused on the post-initialization phases of k-means. We prove that our proposed initialization algorithm k-means|| obtains a nearly optimal solution after a logarithmic number of passes, and then show that in practice a constant number of passes suffices. Experimental evaluation on real-world large-scale data demonstrates that k-means|| outperforms k-means++ in both sequential and parallel settings.

cs.DB↗

arXiv2012

Ad Serving Using a Compact Allocation Plan

A large fraction of online display advertising is sold via guaranteed contracts: a publisher guarantees to the advertiser a certain number of user visits satisfying the targeting predicates of the contract. The publisher is then tasked with solving the ad serving problem - given a user visit, which of the thousands of matching contracts should be displayed, so that by the expiration time every contract has obtained the requisite number of user visits. The challenges of the problem come from (1) the sheer size of the problem being solved, with tens of thousands of contracts and billions of user visits, (2) the unpredictability of user behavior, since these contracts are sold months ahead of time, when only a forecast of user visits is available and (3) the minute amount of resources available online, as an ad server must respond with a matching contract in a fraction of a second. We present a solution to the guaranteed delivery ad serving problem using {\em compact allocation plans}. These plans, computed offline, can be efficiently queried by the ad server during an ad call; they are small, using only O(1) space for contract; and are stateless, allowing for distributed serving without any central coordination. We evaluate this approach on a real set of user visits and guaranteed contracts and show that the compact allocation plans are an effective way of solving the guaranteed delivery ad serving problem.

cs.DS↗

arXiv2012

SHALE: An Efficient Algorithm for Allocation of Guaranteed Display Advertising

Motivated by the problem of optimizing allocation in guaranteed display advertising, we develop an efficient, lightweight method of generating a compact {\em allocation plan} that can be used to guide ad server decisions. The plan itself uses just O(1) state per guaranteed contract, is robust to noise, and allows us to serve (provably) nearly optimally. The optimization method we develop is scalable, with a small in-memory footprint, and working in linear time per iteration. It is also "stop-anytime", meaning that time-critical applications can stop early and still get a good serving solution. Thus, it is particularly useful for optimizing the large problems arising in the context of display advertising. We demonstrate the effectiveness of our algorithm using actual Yahoo! data.

cs.DS↗

arXiv2012

Densest Subgraph in Streaming and MapReduce

The problem of finding locally dense components of a graph is an important primitive in data analysis, with wide-ranging applications from community mining to spam detection and the discovery of biological network modules. In this paper we present new algorithms for finding the densest subgraph in the streaming model. For any epsilon>0, our algorithms make O((log n)/log (1+epsilon)) passes over the input and find a subgraph whose density is guaranteed to be within a factor 2(1+epsilon) of the optimum. Our algorithms are also easily parallelizable and we illustrate this by realizing them in the MapReduce model. In addition we perform extensive experimental evaluation on massive real-world graphs showing the performance and scalability of our algorithms in practice.

cs.DB↗

arXiv2011

Factorization-based Lossless Compression of Inverted Indices

Many large-scale Web applications that require ranked top-k retrieval such as Web search and online advertising are implemented using inverted indices. An inverted index represents a sparse term-document matrix, where non-zero elements indicate the strength of term-document association. In this work, we present an approach for lossless compression of inverted indices. Our approach maps terms in a document corpus to a new term space in order to reduce the number of non-zero elements in the term-document matrix, resulting in a more compact inverted index. We formulate the problem of selecting a new term space that minimizes the resulting index size as a matrix factorization problem, and prove that finding the optimal factorization is an NP-hard problem. We develop a greedy algorithm for finding an approximate solution. A side effect of our approach is increasing the number of terms in the index, which may negatively affect query evaluation performance. To eliminate such effect, we develop a methodology for modifying query evaluation algorithms by exploiting specific properties of our compression approach. Our experimental evaluation demonstrates that our approach achieves an index size reduction of 20%, while maintaining the same query response times. Higher compression ratios up to 35% are achievable, however at the cost of slightly longer query response times. Furthermore, combining our approach with other lossless compression techniques, namely variable-byte encoding, leads to index size reduction of up to 50%.

cs.IR↗

arXiv2010

Inventory Allocation for Online Graphical Display Advertising

We discuss a multi-objective/goal programming model for the allocation of inventory of graphical advertisements. The model considers two types of campaigns: guaranteed delivery (GD), which are sold months in advance, and non-guaranteed delivery (NGD), which are sold using real-time auctions. We investigate various advertiser and publisher objectives such as (a) revenue from the sale of impressions, clicks and conversions, (b) future revenue from the sale of NGD inventory, and (c) "fairness" of allocation. While the first two objectives are monetary, the third is not. This combination of demand types and objectives leads to potentially many variations of our model, which we delineate and evaluate. Our experimental results, which are based on optimization runs using real data sets, demonstrate the effectiveness and flexibility of the proposed model.

cs.CE↗

arXiv2009

Social Networks and Stable Matchings in the Job Market

For most people, social contacts play an integral part in finding a new job. As observed by Granovetter's seminal study, the proportion of jobs obtained through social contacts is usually large compared to those obtained through postings or agencies. At the same time, job markets are a natural example of two-sided matching markets. An important solution concept in such markets is that of stable matchings, and the use of the celebrated Gale-Shapley algorithm to compute them. So far, the literature has evolved separately, either focusing on the implications of information flowing through a social network, or on developing a mathematical theory of job markets through the use of two-sided matching techniques. In this paper we provide a model of the job market that brings both aspects of job markets together. To model the social scientists' observations, we assume that workers learn only about positions in firms through social contacts. Given that information structure, we study both static properties of what we call locally stable matchings (i.e., stable matchings subject to informational constraints given by a social network) and dynamic properties through a reinterpretation of Gale-Shapley's algorithm as myopic best response dynamics. We prove that, in general, the set of locally stable matching strictly contains that of stable matchings and it is in fact NP-complete to determine if they are identical. We also show that the lattice structure of stable matchings is in general absent. Finally, we focus on myopic best response dynamics inspired by the Gale-Shapley algorithm. We study the efficiency loss due to the informational constraints, providing both lower and upper bounds.

cs.GT↗

arXiv2009

Bidding for Representative Allocations for Display Advertising

Display advertising has traditionally been sold via guaranteed contracts -- a guaranteed contract is a deal between a publisher and an advertiser to allocate a certain number of impressions over a certain period, for a pre-specified price per impression. However, as spot markets for display ads, such as the RightMedia Exchange, have grown in prominence, the selection of advertisements to show on a given page is increasingly being chosen based on price, using an auction. As the number of participants in the exchange grows, the price of an impressions becomes a signal of its value. This correlation between price and value means that a seller implementing the contract through bidding should offer the contract buyer a range of prices, and not just the cheapest impressions necessary to fulfill its demand. Implementing a contract using a range of prices, is akin to creating a mutual fund of advertising impressions, and requires {\em randomized bidding}. We characterize what allocations can be implemented with randomized bidding, namely those where the desired share obtained at each price is a non-increasing function of price. In addition, we provide a full characterization of when a set of campaigns are compatible and how to implement them with randomized bidding strategies.

cs.MA↗