arXiv subjects

Rasmus Pagh

Publications and source records attributed to Rasmus Pagh.

At least 91 records · Page 5Linked to original sources

arXiv2014

Approximate Range Emptiness in Constant Time and Optimal Space

This paper studies the \emph{$\varepsilon$-approximate range emptiness} problem, where the task is to represent a set $S$ of $n$ points from $\{0,\ldots,U-1\}$ and answer emptiness queries of the form "$[a ; b]\cap S \neq \emptyset$ ?" with a probability of \emph{false positives} allowed. This generalizes the functionality of \emph{Bloom filters} from single point queries to any interval length $L$. Setting the false positive rate to $\varepsilon/L$ and performing $L$ queries, Bloom filters yield a solution to this problem with space $O(n \lg(L/\varepsilon))$ bits, false positive probability bounded by $\varepsilon$ for intervals of length up to $L$, using query time $O(L \lg(L/\varepsilon))$. Our first contribution is to show that the space/error trade-off cannot be improved asymptotically: Any data structure for answering approximate range emptiness queries on intervals of length up to $L$ with false positive probability $\varepsilon$, must use space $Ω(n \lg(L/\varepsilon)) - O(n)$ bits. On the positive side we show that the query time can be improved greatly, to constant time, while matching our space lower bound up to a lower order additive term. This result is achieved through a succinct data structure for (non-approximate 1d) range emptiness/reporting queries, which may be of independent interest.

cs.DS↗

arXiv2014

Consistent Subset Sampling

Consistent sampling is a technique for specifying, in small space, a subset $S$ of a potentially large universe $U$ such that the elements in $S$ satisfy a suitably chosen sampling condition. Given a subset $\mathcal{I}\subseteq U$ it should be possible to quickly compute $\mathcal{I}\cap S$, i.e., the elements in $\mathcal{I}$ satisfying the sampling condition. Consistent sampling has important applications in similarity estimation, and estimation of the number of distinct items in a data stream. In this paper we generalize consistent sampling to the setting where we are interested in sampling size-$k$ subsets occurring in some set in a collection of sets of bounded size $b$, where $k$ is a small integer. This can be done by applying standard consistent sampling to the $k$-subsets of each set, but that approach requires time $Θ(b^k)$. Using a carefully designed hash function, for a given sampling probability $p \in (0,1]$, we show how to improve the time complexity to $Θ(b^{\lceil k/2\rceil}\log \log b + pb^k)$ in expectation, while maintaining strong concentration bounds for the sample. The space usage of our method is $Θ(b^{\lceil k/4\rceil})$. We demonstrate the utility of our technique by applying it to several well-studied data mining problems. We show how to efficiently estimate the number of frequent $k$-itemsets in a stream of transactions and the number of bipartite cliques in a graph given as incidence stream. Further, building upon a recent work by Campagna et al., we show that our approach can be applied to frequent itemset mining in a parallel or distributed setting. We also present applications in graph stream mining.

cs.DS↗

arXiv2014

The Input/Output Complexity of Triangle Enumeration

We consider the well-known problem of enumerating all triangles of an undirected graph. Our focus is on determining the input/output (I/O) complexity of this problem. Let $E$ be the number of edges, $M 0$. Our results are based on a new color coding technique, which may be of independent interest.

cs.DS↗

arXiv2014

The Input/Output Complexity of Sparse Matrix Multiplication

We consider the problem of multiplying sparse matrices (over a semiring) where the number of non-zero entries is larger than main memory. In the classical paper of Hong and Kung (STOC '81) it was shown that to compute a product of dense $U \times U$ matrices, $Θ\left(U^3 / (B \sqrt{M}) \right)$ I/Os are necessary and sufficient in the I/O model with internal memory size $M$ and memory block size $B$. In this paper we generalize the upper and lower bounds of Hong and Kung to the sparse case. Our bounds depend of the number $N = \mathtt{nnz}(A)+\mathtt{nnz}(C)$ of nonzero entries in $A$ and $C$, as well as the number $Z = \mathtt{nnz}(AC)$ of nonzero entries in $AC$. We show that $AC$ can be computed using $\tilde{O} \left(\tfrac{N}{B} \min\left(\sqrt{\tfrac{Z}{M}},\tfrac{N}{M}\right) \right)$ I/Os, with high probability. This is tight (up to polylogarithmic factors) when only semiring operations are allowed, even for dense rectangular matrices: We show a lower bound of $Ω\left(\tfrac{N}{B} \min\left(\sqrt{\tfrac{Z}{M}},\tfrac{N}{M}\right) \right)$ I/Os. While our lower bound uses fairly standard techniques, the upper bound makes use of ``compressed matrix multiplication'' sketches, which is new in the context of I/O-efficient algorithms, and a new matrix product size estimation technique that avoids the ``no cancellation'' assumption.

cs.DS↗

arXiv2013

How to Approximate A Set Without Knowing Its Size In Advance

The dynamic approximate membership problem asks to represent a set S of size n, whose elements are provided in an on-line fashion, supporting membership queries without false negatives and with a false positive rate at most epsilon. That is, the membership algorithm must be correct on each x in S, and may err with probability at most epsilon on each x not in S. We study a well-motivated, yet insufficiently explored, variant of this problem where the size n of the set is not known in advance. Existing optimal approximate membership data structures require that the size is known in advance, but in many practical scenarios this is not a realistic assumption. Moreover, even if the eventual size n of the set is known in advance, it is desirable to have the smallest possible space usage also when the current number of inserted elements is smaller than n. Our contribution consists of the following results: - We show a super-linear gap between the space complexity when the size is known in advance and the space complexity when the size is not known in advance. - We show that our space lower bound is tight, and can even be matched by a highly efficient data structure.

cs.DS↗

arXiv2012

On Parallelizing Matrix Multiplication by the Column-Row Method

We consider the problem of sparse matrix multiplication by the column row method in a distributed setting where the matrix product is not necessarily sparse. We present a surprisingly simple method for "consistent" parallel processing of sparse outer products (column-row vector products) over several processors, in a communication-avoiding setting where each processor has a copy of the input. The method is consistent in the sense that a given output entry is always assigned to the same processor independently of the specific structure of the outer product. We show guarantees on the work done by each processor, and achieve linear speedup down to the point where the cost is dominated by reading the input. Our method gives a way of distributing (or parallelizing) matrix product computations in settings where the main bottlenecks are storing the result matrix, and inter-processor communication. Motivated by observations on real data that often the absolute values of the entries in the product adhere to a power law, we combine our approach with frequent items mining algorithms and show how to obtain a tight approximation of the weight of the heaviest entries in the product matrix. As a case study we present the application of our approach to frequent pair mining in transactional data streams, a problem that can be phrased in terms of sparse ${0,1}$-integer matrix multiplication by the column-row method. Experimental evaluation of the proposed method on real-life data supports the theoretical findings.

cs.DS↗

arXiv2012

Thresholds for Extreme Orientability

Multiple-choice load balancing has been a topic of intense study since the seminal paper of Azar, Broder, Karlin, and Upfal. Questions in this area can be phrased in terms of orientations of a graph, or more generally a k-uniform random hypergraph. A (d,b)-orientation is an assignment of each edge to d of its vertices, such that no vertex has more than b edges assigned to it. Conditions for the existence of such orientations have been completely documented except for the "extreme" case of (k-1,1)-orientations. We consider this remaining case, and establish: - The density threshold below which an orientation exists with high probability, and above which it does not exist with high probability. - An algorithm for finding an orientation that runs in linear time with high probability, with explicit polynomial bounds on the failure probability. Previously, the only known algorithms for constructing (k-1,1)-orientations worked for k<=3, and were only shown to have expected linear running time.

cs.DS↗

arXiv2011

Compressed Matrix Multiplication

Motivated by the problems of computing sample covariance matrices, and of transforming a collection of vectors to a basis where they are sparse, we present a simple algorithm that computes an approximation of the product of two n-by-n real matrices A and B. Let ||AB||_F denote the Frobenius norm of AB, and b be a parameter determining the time/accuracy trade-off. Given 2-wise independent hash functions $_1,h_2: [n] -> [b], and s_1,s_2: [n] -> {-1,+1} the algorithm works by first "compressing" the matrix product into the polynomial p(x) = sum_{k=1}^n (sum_{i=1}^n A_{ik} s_1(i) x^{h_1(i)}) (sum_{j=1}^n B_{kj} s_2(j) x^{h_2(j)}) Using FFT for polynomial multiplication, we can compute c_0,...,c_{b-1} such that sum_i c_i x^i = (p(x) mod x^b) + (p(x) div x^b) in time Õ(n^2+ n b). An unbiased estimator of (AB)_{ij} with variance at most ||AB||_F^2 / b can then be computed as: C_{ij} = s_1(i) s_2(j) c_{(h_1(i)+h_2(j)) mod b. Our approach also leads to an algorithm for computing AB exactly, whp., in time Õ(N + nb) in the case where A and B have at most N nonzero entries, and AB has at most b nonzero entries. Also, we use error-correcting codes in a novel way to recover significant entries of AB in near-linear time.

cs.DS↗

arXiv2011

I/O-Efficient Data Structures for Colored Range and Prefix Reporting

Motivated by information retrieval applications, we consider the one-dimensional colored range reporting problem in rank space. The goal is to build a static data structure for sets C_1,...,C_m \subseteq {1,...,sigma} that supports queries of the kind: Given indices a,b, report the set Union_{a <= i <= b} C_i. We study the problem in the I/O model, and show that there exists an optimal linear-space data structure that answers queries in O(1+k/B) I/Os, where k denotes the output size and B the disk block size in words. In fact, we obtain the same bound for the harder problem of three-sided orthogonal range reporting. In this problem, we are to preprocess a set of n two-dimensional points in rank space, such that all points inside a query rectangle of the form [x_1,x_2] x (-infinity,y] can be reported. The best previous bounds for this problem is either O(n lg^2_B n) space and O(1+k/B) query I/Os, or O(n) space and O(lg^(h)_B n +k/B) query I/Os, where lg^(h)_B n is the base B logarithm iterated h times, for any constant integer h. The previous bounds are both achieved under the indivisibility assumption, while our solution exploits the full capabilities of the underlying machine. Breaking the indivisibility assumption thus provides us with cleaner and optimal bounds. Our results also imply an optimal solution to the following colored prefix reporting problem. Given a set S of strings, each O(1) disk blocks in length, and a function c: S -> 2^{1,...,sigma}, support queries of the kind: Given a string p, report the set Union_{x in S intersection p*} c(x), where p* denotes the set of strings with prefix p. Finally, we consider the possibility of top-k extensions of this result, and present a simple solution in a model that allows non-blocked I/O.

cs.DS↗

arXiv2011

Colorful Triangle Counting and a MapReduce Implementation

In this note we introduce a new randomized algorithm for counting triangles in graphs. We show that under mild conditions, the estimate of our algorithm is strongly concentrated around the true number of triangles. Specifically, if $p \geq \max{(\frac{Δ\log{n}}{t}, \frac{\log{n}}{\sqrt{t}})}$, where $n$, $t$, $Δ$ denote the number of vertices in $G$, the number of triangles in $G$, the maximum number of triangles an edge of $G$ is contained, then for any constant $ε>0$ our unbiased estimate $T$ is concentrated around its expectation, i.e., $ \Prob{|T - \Mean{T}| \geq ε\Mean{T}} = o(1)$. Finally, we present a \textsc{MapReduce} implementation of our algorithm.

cs.DS↗

arXiv2011

Better size estimation for sparse matrix products

We consider the problem of doing fast and reliable estimation of the number of non-zero entries in a sparse boolean matrix product. This problem has applications in databases and computer algebra. Let n denote the total number of non-zero entries in the input matrices. We show how to compute a 1 +- epsilon approximation (with small probability of error) in expected time O(n) for any epsilon > 4/\sqrt[4]{z}. The previously best estimation algorithm, due to Cohen (JCSS 1997), uses time O(n/epsilon^2). We also present a variant using O(sort(n)) I/Os in expectation in the cache-oblivious model. In contrast to these results, the currently best algorithms for computing a sparse boolean matrix product use time omega(n^{4/3}) (resp. omega(n^{4/3}/B) I/Os), even if the result matrix has only z=O(n) nonzero entries. Our algorithm combines the size estimation technique of Bar-Yossef et al. (RANDOM 2002) with a particular class of pairwise independent hash functions that allows the sketch of a set of the form A x C to be computed in expected time O(|A|+|C|) and O(sort(|A|+|C|)) I/Os. We then describe how sampling can be used to maintain (independent) sketches of matrices that allow estimation to be performed in time o(n) if z is sufficiently large. This gives a simpler alternative to the sketching technique of Ganguly et al. (PODS 2005), and matches a space lower bound shown in that paper. Finally, we present experiments on real-world data sets that show the accuracy of both our methods to be significantly better than the worst-case analysis predicts.

cs.DS↗

arXiv2011

A New Data Layout For Set Intersection on GPUs

Set intersection is the core in a variety of problems, e.g. frequent itemset mining and sparse boolean matrix multiplication. It is well-known that large speed gains can, for some computational problems, be obtained by using a graphics processing unit (GPU) as a massively parallel computing device. However, GPUs require highly regular control flow and memory access patterns, and for this reason previous GPU methods for intersecting sets have used a simple bitmap representation. This representation requires excessive space on sparse data sets. In this paper we present a novel data layout, "BatMap", that is particularly well suited for parallel processing, and is compact even for sparse data. Frequent itemset mining is one of the most important applications of set intersection. As a case-study on the potential of BatMaps we focus on frequent pair mining, which is a core special case of frequent itemset mining. The main finding is that our method is able to achieve speedups over both Apriori and FP-growth when the number of distinct items is large, and the density of the problem instance is above 1%. Previous implementations of frequent itemset mining on GPU have not been able to show speedups over the best single-threaded implementations.

cs.DS↗

arXiv2010

Tight Thresholds for Cuckoo Hashing via XORSAT

We settle the question of tight thresholds for offline cuckoo hashing. The problem can be stated as follows: we have n keys to be hashed into m buckets each capable of holding a single key. Each key has k >= 3 (distinct) associated buckets chosen uniformly at random and independently of the choices of other keys. A hash table can be constructed successfully if each key can be placed into one of its buckets. We seek thresholds alpha_k such that, as n goes to infinity, if n/m <= alpha for some alpha < alpha_k then a hash table can be constructed successfully with high probability, and if n/m >= alpha for some alpha > alpha_k a hash table cannot be constructed successfully with high probability. Here we are considering the offline version of the problem, where all keys and hash values are given, so the problem is equivalent to previous models of multiple-choice hashing. We find the thresholds for all values of k > 2 by showing that they are in fact the same as the previously known thresholds for the random k-XORSAT problem. We then extend these results to the setting where keys can have differing number of choices, and provide evidence in the form of an algorithm for a conjecture extending this result to cuckoo hash tables that store multiple keys in a bucket.

cs.DS↗

arXiv2010

On Finding Frequent Patterns in Event Sequences

Given a directed acyclic graph with labeled vertices, we consider the problem of finding the most common label sequences ("traces") among all paths in the graph (of some maximum length m). Since the number of paths can be huge, we propose novel algorithms whose time complexity depends only on the size of the graph, and on the frequency epsilon of the most frequent traces. In addition, we apply techniques from streaming algorithms to achieve space usage that depends only on epsilon, and not on the number of distinct traces. The abstract problem considered models a variety of tasks concerning finding frequent patterns in event sequences. Our motivation comes from working with a data set of 2 million RFID readings from baggage trolleys at Copenhagen Airport. The question of finding frequent passenger movement patterns is mapped to the above problem. We report on experimental findings for this data set.

cs.DS↗

arXiv2010

On Finding Similar Items in a Stream of Transactions

While there has been a lot of work on finding frequent itemsets in transaction data streams, none of these solve the problem of finding similar pairs according to standard similarity measures. This paper is a first attempt at dealing with this, arguably more important, problem. We start out with a negative result that also explains the lack of theoretical upper bounds on the space usage of data mining algorithms for finding frequent itemsets: Any algorithm that (even only approximately and with a chance of error) finds the most frequent k-itemset must use space Omega(min{mb,n^k,(mb/phi)^k}) bits, where mb is the number of items in the stream so far, n is the number of distinct items and phi is a support threshold. To achieve any non-trivial space upper bound we must thus abandon a worst-case assumption on the data stream. We work under the model that the transactions come in random order, and show that surprisingly, not only is small-space similarity mining possible for the most common similarity measures, but the mining accuracy improves with the length of the stream for any fixed support threshold.

cs.DS↗

arXiv2010

On Finding Frequent Patterns in Directed Acyclic Graphs

Given a directed acyclic graph with labeled vertices, we consider the problem of finding the most common label sequences ("traces") among all paths in the graph (of some maximum length m). Since the number of paths can be huge, we propose novel algorithms whose time complexity depends only on the size of the graph, and on the relative frequency epsilon of the most frequent traces. In addition, we apply techniques from streaming algorithms to achieve space usage that depends only on epsilon, and not on the number of distinct traces. The abstract problem considered models a variety of tasks concerning finding frequent patterns in event sequences. Our motivation comes from working with a data set of 2 million RFID readings from baggage trolleys at Copenhagen Airport. The question of finding frequent passenger movement patterns is mapped to the above problem. We report on experimental findings for this data set.

cs.DS↗

arXiv2010

Finding Associations and Computing Similarity via Biased Pair Sampling

This version is ***superseded*** by a full version that can be found at http://www.itu.dk/people/pagh/papers/mining-jour.pdf, which contains stronger theoretical results and fixes a mistake in the reporting of experiments. Abstract: Sampling-based methods have previously been proposed for the problem of finding interesting associations in data, even for low-support items. While these methods do not guarantee precise results, they can be vastly more efficient than approaches that rely on exact counting. However, for many similarity measures no such methods have been known. In this paper we show how a wide variety of measures can be supported by a simple biased sampling method. The method also extends to find high-confidence association rules. We demonstrate theoretically that our method is superior to exact methods when the threshold for "interesting similarity/confidence" is above the average pairwise similarity/confidence, and the average support is not too low. Our method is particularly good when transactions contain many items. We confirm in experiments on standard association mining benchmarks that this gives a significant speedup on real data sets (sometimes much larger than the theoretical guarantees). Reductions in computation time of over an order of magnitude, and significant savings in space, are observed.

cs.DS↗

arXiv2008

Secondary Indexing in One Dimension: Beyond B-trees and Bitmap Indexes

Let S be a finite, ordered alphabet, and let x = x_1 x_2 ... x_n be a string over S. A "secondary index" for x answers alphabet range queries of the form: Given a range [a_l,a_r] over S, return the set I_{[a_l;a_r]} = {i |x_i \in [a_l; a_r]}. Secondary indexes are heavily used in relational databases and scientific data analysis. It is well-known that the obvious solution, storing a dictionary for the position set associated with each character, does not always give optimal query time. In this paper we give the first theoretically optimal data structure for the secondary indexing problem. In the I/O model, the amount of data read when answering a query is within a constant factor of the minimum space needed to represent I_{[a_l;a_r]}, assuming that the size of internal memory is (|S| log n)^{delta} blocks, for some constant delta > 0. The space usage of the data structure is O(n log |S|) bits in the worst case, and we further show how to bound the size of the data structure in terms of the 0-th order entropy of x. We show how to support updates achieving various time-space trade-offs. We also consider an approximate version of the basic secondary indexing problem where a query reports a superset of I_{[a_l;a_r]} containing each element not in I_{[a_l;a_r]} with probability at most epsilon, where epsilon > 0 is the false positive probability. For this problem the amount of data that needs to be read by the query algorithm is reduced to O(|I_{[a_l;a_r]}| log(1/epsilon)) bits.

cs.DB↗