We typically have seminars on Wednesday at noon in Malone 228. All seminar announcements will be sent to the theory mailing list.

Speaker: Karthik Abinav Sankararaman

Affiliation: University of Maryland

Title: Adversarial Bandits with Knapsacks

Abstract: In this talk we will discuss the multi-armed bandits problem with resource constraints under the adversarial setting. In this problem, we have an interactive and repeated game between the algorithm and an adversary. Given T time-steps, d resources, m actions and budgets B1, B2, .. Bd, the algorithm chooses one of the m actions at each time-step. An adversary then reveals a reward and consumption for each of the d resources corresponding to this action. The time-step at which the algorithm runs out of the d resources (i.e., the total consumption for resource j > Bj), the game stops and the total reward is the sum of rewards obtained until the stopping time. The goal is to maximize the competitive ratio; the ratio of the total reward of the algorithm to the expected reward of a fixed distribution that knows all the rewards and consumption ahead of time. We give an algorithm for this problem whose competitive ratio is tight (matches the lower-bound). Moreover the algorithmic tools extends in an (almost) black-box fashion to also give an algorithm for the stochastic setting thus giving a “best-of-both-worlds” algorithm where the algorithm need not know a-priori if the input is adversarial or i.i.d. Finally we conclude with applications and special cases including the Dynamic Pricing problem.

This talk is based on a recent working paper with Nicole Immorlica, Rob Schapire and Alex Slivkins.

Speaker: Nithin Varma

Affiliation: Boston University

Title: Separating erasures and errors in property testing using local list decoding

Abstract:

Corruption in data can be in the form of erasures (missing data) or errors (wrong data). Erasure-resilient property testing (Dixit, Raskhodnikova, Thakurta, Varma ’16) and tolerant property testing (Parnas, Ron, Rubinfeld ’06) are two formal models of sublinear algorithms that account for the presence of erasures and errors in input data, respectively.

We first show that there exists a property P that has an erasure-resilient tester whose query complexity is independent of the input size n, whereas every tolerant tester for P has query complexity that depends on n. We obtain this result by designing a local list decoder for the Hadamard code that works in the presence of erasures, thereby proving an analog of the famous Goldreich-Levin Theorem. We also show a strengthened separation by proving that there exists another property R such that R has a constant-query erasure-resilient tester, whereas every tolerant tester for R requires n^{Omega(1)} queries. The main tool used in proving the strengthened separation is an approximate variant of a locally list decodable code that works against erasures.

Joint work with Sofya Raskhodnikova and Noga Ron-Zewi.

Speaker: Jalaj Upadhyay

Affiliation: JHU

Title: Differentially Private Spectral Sparsification of Graphs

Abstract:

In this talk, we will discuss differentially private spectral sparsification of graphs. We argue that traditional spectral sparsification where the output graph is a subgraph of the input graph is not possible with differential privacy. This motivates us to define a relaxed version of spectral sparsification of graphs.

We consider edge-level privacy, i.e., neighboring graphs differs in one edge with weight one. We give efficient $(\alpha,\beta)$-differentially private algorithms that, on input a dense graph G, construct a spectral sparsification of G. Our output graphs has $ O(n/\eps^2)$ weighted edges, which matches the best known non-private algorithms.

We can use our private sparse graph to solve various combinatorial and learning problems on graphs efficiently while preserving differential privacy. Some examples include all possible cut queries, Max-Cut, Sparse-Cut, Edge-Expansion, Laplacian eigenmaps, etc.

This talk is based on a joint work with Raman Arora and Vladimir Braverman.

Speaker: Ke Wu

Affiliation: Johns Hopkins University

Title: Synchronization Strings: Efficient and Fast Deterministic Constructions over Small Alphabets

Abstract:

Synchronization strings are recently introduced by Haeupler and Shahrasbi (STOC 2017) in the study of codes for correcting insertion and deletion errors (insdel codes). They showed that for any parameter ε>0, synchronization strings of arbitrary length exist over an alphabet whose size depends only on ε. Specifically, they obtained an alphabet size of O(ε^{−4}), which left an open question on where the minimal size of such alphabets lies between Ω(ε^{1}) and O(ε^{−4}). In this work, we partially bridge this gap by providing an improved lower bound of Ω(ε^{−3/2}), and an improved upper bound of O(ε^{−2}). We also provide fast explicit constructions of synchronization strings over small alphabets.

Further, along the lines of previous work on similar combinatorial objects, we study the extremal question of the smallest possible alphabet size over which synchronization strings can exist for some constant ε<1. We show that one can construct ε-synchronization strings over alphabets of size four while no such string exists over binary alphabets. This reduces the extremal question to whether synchronization strings exist over ternary alphabets.

Speaker: Sami Davies

Affiliation: University of Washington

Title: A Tale of Santa Claus, Hypergraphs, and Matroids

Abstract:

A well-known problem in scheduling and approximation algorithms is the Santa Claus problem. Suppose that Santa Claus has a set of gifts, and he wants to distribute them among a set of children so that the least happy child is made as happy as possible. Here, the value that a child i has for a present j is of the form p_{ij} \in \{0,p_j\}. A polynomial time algorithm by Annamalai et al. gives a 12.33-approximation algorithm and is based on a modification of Haxell’s hypergraph matching argument.

In this paper, we introduce a matroid version of the Santa Claus problem. Our algorithm is also based on Haxell’s augmentation tree, but with the introduction of the matroid structure we solve a more general problem with cleaner methods. Our result can then be used as a blackbox to obtain a (4 +\varepsilon)-approximation for Santa Claus. This factor also compares against a natural, compact LP for Santa Claus.

Speaker: Martin Farach-Colton

Affiliation: Rutgers University

Title: TBA

Abstract: TBA