Feature hashing algorithms [14], [21] are well-known algorithmic techniques for handling large dimensionality of the datasets. These methods reduce the dimensionality of data points while preserving the pairwise distance, or similarity between them. However, to the best of our knowledge, these feature hashing algorithms are not adaptable to the dynamic insertion and deletion of features. In this work, we suggest algorithms using which the existing feature hashing algorithms can be made adapted to dynamic feature insertion and deletion. Our algorithms fit in the framework of both real-valued [21], and binary [14] feature hashing algorithms. We show a theoretical analysis of our algorithms, and complement it with experiments on real-world datasets. Our algorithms suggest comparable accuracy w.r.t. baselines while simultaneously offering a significant speed-up in the running time. Our proposal is easy to implement and can be adopted in practice.

Generating succinct summaries (also known as sketches) of massive data streams is becoming increasingly important. Such a task typically requires fast, accurate, and small space algorithms in order to support the downstream applications, mainly in areas such as data analysis, machine learning and data mining. A fundamental and well-studied problem in this context is that of estimating the frequencies of the items appearing in a data stream. The Count-Min-Sketch (Cormode and Muthukrishnan, J. Algorithms, 55(1):58–75, 2005) and Count-Sketch (Charikar et al., Theor. Comput. Sci., 312(1):3–15, 2004) are two known classical algorithms for this purpose. However, a limitation of these techniques is that the variance of their estimate tends to be large. In this work, we address this problem and suggest a technique that reduces the variance in their respective estimates, at the cost of little computational overhead. Our technique relies on the classical Control-Variate trick (Lavenberg and Welch, Manage. Sci., 27:322–335, 1981) used for reducing variance in Monte-Carlo simulation. We present a theoretical analysis of our proposal by carefully choosing the control variates and complement them with experiments on synthetic as well as real-world datasets.

We present sketching algorithms for sparse binary datasets, which maintain binary version of the dataset after sketching, while simultaneously preserving multiple similarity measures such as Jaccard Similarity, Cosine Similarity, Inner Product, and Hamming Distance, on the same sketch. A major advantage of our algorithms is that they are randomness efficient, and require significantly less number of random bits for sketching – logarithmic in dimension, while other competitive algorithms require linear in dimension. Our proposed algorithms are efficient, offer a compact sketch of the dataset, and can be efficiently deployed in a distributive setting. We present a theoretical analysis of our approach and complement them with extensive experimentations on public datasets. For analysis purposes, our algorithms require a natural assumption on the dataset. We empirically verify the assumption and notice that it holds on several real-world datasets.

We propose a dimensionality reduction (sketching) algorithm for high dimensional, sparse, binary data. Our proposed algorithm provides a single sketch which simultaneously preserves multiple similarity measures including Hamming distance, Inner product, and Jaccard Similarity [12]. In contrast to the "local projection" strategy used by most of the earlier algorithms [6], [4], [7], our approach exploits sparsity and combines the following two strategies: 1. partitioning the dimensions into several buckets, 2. obtaining "global linear summaries" within those buckets. Our algorithm is faster than the existing state-of-the-art, and it preserves the binary format of the data after the dimensionality reduction, which makes the sketch space efficient. Our algorithm can also be easily adapted in streaming and incremental learning frameworks. We give a rigorous theoretical analysis of the dimensionality reduction bounds and complement it with extensive experiments. Our proposed algorithm is simple and easy to implement in practice.

The well-known k-disjoint path problem (k-DPP) asks for pairwise vertex-disjoint paths between k specified pairs of vertices (s_i, t_i) in a given graph, if they exist. The decision version of the shortest k-DPP asks for the length of the shortest (in terms of total length) such paths. Similarly, the search and counting versions ask for one such and the number of such shortest set of paths, respectively. We restrict attention to the shortest k-DPP instances on undirected planar graphs where all sources and sinks lie on a single face or on a pair of faces. We provide efficient sequential and parallel algorithms for the search versions of the problem answering one of the main open questions raised by Colin de Verdière and Schrijver [Éric Colin de Verdière and Alexander Schrijver, 2011] for the general one-face problem. We do so by providing a randomised NC^2 algorithm along with an O(n^{omega/2}) time randomised sequential algorithm, for any fixed k. We also obtain deterministic algorithms with similar resource bounds for the counting and search versions. In contrast, previously, only the sequential complexity of decision and search versions of the "well-ordered" case has been studied. For the one-face case, sequential versions of our routines have better running times for constantly many terminals. The algorithms are based on a bijection between a shortest k-tuple of disjoint paths in the given graph and cycle covers in a related digraph. This allows us to non-trivially modify established techniques relating counting cycle covers to the determinant. We further need to do a controlled inclusion-exclusion to produce a polynomial sum of determinants such that all "bad" cycle covers cancel out in the sum allowing us to count "pure" cycle covers.

Perfect matchings in planar graphs have been extensively studied and understood in the context of parallel complexity [P W Kastelyn, 1967; Vijay Vazirani, 1988; Meena Mahajan and Kasturi R. Varadarajan, 2000; Datta et al., 2010; Nima Anari and Vijay V. Vazirani, 2017]. However, corresponding results for maximum matchings have been elusive. We partly bridge this gap by proving: 1) An SPL upper bound for planar bipartite maximum matching search. 2) Planar maximum matching search reduces to planar maximum matching decision. 3) Planar maximum matching count reduces to planar bipartite maximum matching count and planar maximum matching decision. The first bound improves on the known [Thanh Minh Hoang, 2010] bound of L^{C_=L} and is adaptable to any special bipartite graph class with non-zero circulation such as bounded genus graphs, K_{3,3}-free graphs and K_5-free graphs. Our bounds and reductions non-trivially combine techniques like the Gallai-Edmonds decomposition [L. Lovász and M.D. Plummer, 1986], deterministic isolation [Datta et al., 2010; Samir Datta et al., 2012; Rahul Arora et al., 2016], and the recent breakthroughs in the parallel search for planar perfect matchings [Nima Anari and Vijay V. Vazirani, 2017; Piotr Sankowski, 2018].

Patnaik and Immerman introduced the dynamic complexity class DynFO of database queries that can be maintained by first-order dynamic programs with the help of auxiliary relations under insertions and deletions of edges. This article confirms their conjecture that the reachability query is in DynFO. As a byproduct, it is shown that the rank of a matrix with small values can be maintained in DynFO. It is further shown that the (size of the) maximum matching of a graph can be maintained in non-uniform DynFO, an extension of DynFO, with non-uniform initialisation of the auxiliary relations.

In 2013 Bei, Chen and Zhang introduced a trial and error model of computing, and applied to some constraint satisfaction problems. In this model the input is hidden by an oracle which, for a candidate assignment, reveals some information about a violated constraint if the assignment is not satisfying. In this paper we initiate a systematic study of constraint satisfaction problems in the trial and error model, by adopting a formal framework for CSPs, and defining several types of revealing oracles. Our main contribution is to develop a transfer theorem for each type of the revealing oracle. To any hidden CSP with a specific type of revealing oracle, the transfer theorem associates another CSP in the normal setting, such that their complexities are polynomial-time equivalent. This in principle transfers the study of a large class of hidden CSPs to the study of normal CSPs. We apply the transfer theorems to get polynomial-time algorithms or hardness results for several families of concrete problems.

The query complexity of graph properties is well-studied when queries are on edges. We investigate the same when queries are on nodes. In this setting a graph $G = (V, E)$ on $n$ vertices and a property $\mathcal{P}$ are given. A black-box access to an unknown subset $S \subseteq V$ is provided via queries of the form `Does $i \in S$?'. We are interested in the minimum number of queries needed in worst case in order to determine whether $G[S]$, the subgraph of $G$ induced on $S$, satisfies $\mathcal{P}$. Apart from being combinatorially rich, this setting allows us to initiate a systematic study of breaking symmetry in the context of query complexity of graph properties. In particular, we focus on hereditary graph properties. The monotone functions in the node-query setting translate precisely to the hereditary graph properties. The famous Evasiveness Conjecture asserts that even with a minimal symmetry assumption on $G$, namely that of vertex-transitivity, the query complexity for any hereditary graph property in our setting is the worst possible, i.e., $n$. We show that in the absence of any symmetry on $G$ it can fall as low as $O(n^{1/(d + 1) })$ where $d$ denotes the minimum possible degree of a minimal forbidden sub-graph for $\mathcal{P}$. In particular, every hereditary property benefits at least quadratically. The main question left open is: can it go exponentially low for some hereditary property? We show that the answer is no for any hereditary property with {finitely many} forbidden subgraphs by exhibiting a bound of $Ω(n^{1/k})$ for some constant $k$ depending only on the property. For general ones we rule out the possibility of the query complexity falling down to constant by showing $Ω(\log n/ \log \log n)$ bound. Interestingly, our lower bound proofs rely on the famous Sunflower Lemma due to Erdös and Rado.

We present a Logspace Approximation Scheme (LSAS), i.e. an approximation algorithm for maximum matching in planar graphs (not necessarily bipartite) that achieves an approximation ratio arbitrarily close to one, using only logarithmic space. This deviates from the well known Baker's approach for approximation in planar graphs by avoiding the use of distance computation - which is not known to be in Logspace. Our algorithm actually works for any "recursively sparse" graph class which contains a linear size matching and also for certain other classes like bounded genus graphs. The scheme is based on an LSAS in bounded degree graphs which are not known to be amenable to Baker's method. We solve the bounded degree case by parallel augmentation of short augmenting paths. Finding a large number of such disjoint paths can, in turn, be reduced to finding a large independent set in a bounded degree graph. The bounded degree assumption allows us to obtain a Logspace algorithm.

Let H be a (non-empty) graph on n vertices, possibly containing isolated vertices. Let f_H(G) = 1 iff the input graph G on n vertices contains H as a (not necessarily induced) subgraph. Let alpha_H denote the cardinality of a maximum independent set of H. In this paper we show: Q(f_H) = Omega( sqrt{alpha_H * n}), where Q(f_H) denotes the quantum query complexity of f_H. As a consequence we obtain lower bounds for Q(f_H) in terms of several other parameters of H such as the average degree, minimum vertex cover, chromatic number, and the critical probability. We also use the above bound to show that Q(f_H) = Omega(n^{3/4}) for any H, improving on the previously best known bound of Omega(n^{2/3}) [M. Santha/A. Chi-Chih Yao, unpublished manuscript]. Until very recently, it was believed that the quantum query complexity is at least square root of the randomized one. Our Omega(n^{3/4}) bound for Q(f_H) matches the square root of the current best known bound for the randomized query complexity of f_H, which is Omega(n^{3/2}) due to Groger. Interestingly, the randomized bound of Omega(alpha_H * n) for f_H still remains open. We also study the Subgraph Homomorphism Problem, denoted by f_{[H]}, and show that Q(f_{[H]}) = Omega(n). Finally we extend our results to the 3-uniform hypergraphs. In particular, we show an Omega(n^{4/5}) bound for quantum query complexity of the Subgraph Isomorphism, improving on the previously known Omega(n^{3/4}) bound. For the Subgraph Homomorphism, we obtain an Omega(n^{3/2}) bound for the same.

Cryptocurrencies like Bitcoin and the more recent Ethereum system allow users to specify scripts in transactions and contracts to support applications beyond simple cash transactions. In this work, we analyze the extent to which these systems can enforce the correct semantics of scripts. We show that when a script execution requires nontrivial computation effort, practical attacks exist which either waste miners' computational resources or lead miners to accept incorrect script results. These attacks drive miners to an ill-fated choice, which we call the verifier's dilemma, whereby rational miners are well-incentivized to accept unvalidated blockchains. We call the framework of computation through a scriptable cryptocurrency a consensus computer and develop a model that captures incentives for verifying computation in it. We propose a resolution to the verifier's dilemma which incentivizes correct execution of certain applications, including outsourced computation, where scripts require minimal time to verify. Finally we discuss two distinct, practical implementations of our consensus computer in real cryptocurrency networks like Ethereum.

A function f : {0, 1}n -> {0, 1} is called evasive if its decision tree complexity is maximal, i.e., D(f) = n. The long-standing Anderaa-Rosenberg-Karp (ARK) Conjecture asserts that every non-trivial monotone graph property is evasive. The Evasiveness Conjecture (EC) is a generalization of ARK Conjecture from monotone graph properties to arbitrary monotone transitive Boolean functions.

We exhibit the following new upper bounds on the space complexity and the parallel complexity of the Bipartite Perfect Matching (BPM) problem for graphs of small genus: (1) BPM in planar graphs is in UL (improves upon the SPL bound from Datta, Kulkarni, and Roy; (2) BPM in constant genus graphs is in NL (orthogonal to the SPL bound from Datta, Kulkarni, Tewari, and Vinodchandran.; (3) BPM in poly-logarithmic genus graphs is in NC; (extends the NC bound for O(log n) genus graphs from Mahajan and Varadarajan, and Kulkarni, Mahajan, and Varadarajan. For Part (1) we combine the flow technique of Miller and Naor with the double counting technique of Reinhardt and Allender . For Part (2) and (3) we extend Miller and Naor's result to higher genus surfaces in the spirit of Chambers, Erickson and Nayyeri.

We investigate the space complexity of certain perfect matching problems over bipartite graphs embedded on surfaces of constant genus (orientable or non-orientable). We show that the problems of deciding whether such graphs have (1) a perfect matching or not and (2) a unique perfect matching or not, are in the logspace complexity class \SPL. Since \SPL\ is contained in the logspace counting classes $\oplusŁ$ (in fact in \modk\ for all $k\geq 2$), \CeqL, and \PL, our upper bound places the above-mentioned matching problems in these counting classes as well. We also show that the search version, computing a perfect matching, for this class of graphs is in $\FL^{\SPL}$. Our results extend the same upper bounds for these problems over bipartite planar graphs known earlier. As our main technical result, we design a logspace computable and polynomially bounded weight function which isolates a minimum weight perfect matching in bipartite graphs embedded on surfaces of constant genus. We use results from algebraic topology for proving the correctness of the weight function.

A Boolean function on $N$ variables is called \emph{evasive} if its decision-tree complexity is $N$. A sequence $B_n$ of Boolean functions is \emph{eventually evasive} if $B_n$ is evasive for all sufficiently large $n$. We confirm the eventual evasiveness of several classes of monotone graph properties under widely accepted number theoretic hypotheses. In particular we show that Chowla's conjecture on Dirichlet primes implies that (a) for any graph $H$, ``forbidden subgraph $H$'' is eventually evasive and (b) all nontrivial monotone properties of graphs with $\le n^{3/2-\epsilon}$ edges are eventually evasive. ($n$ is the number of vertices.) While Chowla's conjecture is not known to follow from the Extended Riemann Hypothesis (ERH, the Riemann Hypothesis for Dirichlet's $L$ functions), we show (b) with the bound $O(n^{5/4-\epsilon})$ under ERH. We also prove unconditional results: (a$'$) for any graph $H$, the query complexity of ``forbidden subgraph $H$'' is $\binom{n}{2} - O(1)$; (b$'$) for some constant $c>0$, all nontrivial monotone properties of graphs with $\le cn\log n+O(1)$ edges are eventually evasive. Even these weaker, unconditional results rely on deep results from number theory such as Vinogradov's theorem on the Goldbach conjecture. Our technical contribution consists in connecting the topological framework of Kahn, Saks, and Sturtevant (1984), as further developed by Chakrabarti, Khot, and Shi (2002), with a deeper analysis of the orbital structure of permutation groups and their connection to the distribution of prime numbers. Our unconditional results include stronger versions and generalizations of some result of Chakrabarti et al.

We present a deterministic way of assigning small (log bit) weights to the edges of a bipartite planar graph so that the minimum weight perfect matching becomes unique. The isolation lemma as described in (Mulmuley et al. 1987) achieves the same for general graphs using a randomized weighting scheme, whereas we can do it deterministically when restricted to bipartite planar graphs. As a consequence, we reduce both decision and construction versions of the matching problem to testing whether a matrix is singular, under the promise that its determinant is $0$ or $1$, thus obtaining a highly parallel SPL algorithm for bipartite planar graphs. This improves the earlier known bounds of non-uniform SPL by (Allender et al. 1999) and $NC^2$ by (Miller and Naor 1995, Mahajan and Varadarajan 2000). It also rekindles the hope of obtaining a deterministic parallel algorithm for constructing a perfect matching in non-bipartite planar graphs, which has been open for a long time. Our techniques are elementary and simple.

We consider the problem of counting the number of spanning trees in planar graphs. We prove tight bounds on the complexity of the problem, both in general and especially in the modular setting. We exhibit the problem to be complete for Logspace when the modulus is 2k, for constant k. On the other hand, we show that for any other modulus and in the non-modular case, our problem is as hard in the planar case as for the case of arbitrary graphs. This completely settles the question regarding the complexity of modular computation of the number of spanning trees in planar graphs. The techniques used rely heavily on algebraic-topology. In the spirit of counting problems modulo 2k, we also exhibit a highly parallel oplusL algorithm for finding the value of a Permanent modulo 2k. Previously, the best known result in this direction was Valiant's result that this problem lies in P.