Deeksha Adil, Richard Peng, Sushant Sachdeva
2019-07-16
Linear regression in
-norm is a canonical optimization problem that arises in several applications, including sparse recovery, semi-supervised learning, and signal processing. Generic convex optimization algorithms for solving -regression are slow in practice. Iteratively Reweighted Least Squares (IRLS) is an easy to implement family of algorithms for solving these problems that has been studied for over 50 years. However, these algorithms often diverge for p > 3, and since the work of Osborne (1985), it has been an open problem whether there is an IRLS algorithm that is guaranteed to converge rapidly for p > 3. We propose p-IRLS, the first IRLS algorithm that provably converges geometrically for any Our algorithm is simple to implement and is guaranteed to find a -approximate solution in iterations. Our experiments demonstrate that it performs even better than our theoretical bounds, beats the standard Matlab/CVX implementation for solving these problems by 10--50x, and is the fastest among available implementations in the high-accuracy regime.
Megha Khosla, Avishek Anand
2016-11-23
Hash tables are ubiquitous in computer science for efficient access to large datasets. However, there is always a need for approaches that offer compact memory utilisation without substantial degradation of lookup performance. Cuckoo hashing is an efficient technique of creating hash tables with high space utilisation and offer a guaranteed constant access time. We are given
locations and items. Each item has to be placed in one of the locations chosen by random hash functions. By allowing more than one choice for a single item, cuckoo hashing resembles multiple choice allocations schemes. In addition it supports dynamically changing the location of an item among its possible locations. We propose and analyse an insertion algorithm for cuckoo hashing that runs in \emph{linear time} with high probability and in expectation. Previous work on total allocation time has analysed breadth first search, and it was shown to be linear only in \emph{expectation}. Our algorithm finds an assignment (with probability 1) whenever it exists. In contrast, the other known insertion method, known as \emph{random walk insertion}, may run indefinitely even for a solvable instance. We also present experimental results comparing the performance of our algorithm with the random walk method, also for the case when each location can hold more than one item. As a corollary we obtain a linear time algorithm (with high probability and in expectation) for finding perfect matchings in a special class of sparse random bipartite graphs. We support this by performing experiments on a real world large dataset for finding maximum matchings in general large bipartite graphs. We report an order of magnitude improvement in the running time as compared to the \emph{Hopkraft-Karp} matching algorithm.
Luca Becchetti, Emilio Cruciani, Francesco Pasquale, Sara Rizzo
2019-07-16
Spectral techniques have proved amongst the most effective approaches to graph clustering. However, in general they require explicit computation of the main eigenvectors of a suitable matrix (usually the Laplacian matrix of the graph). Recent work (e.g., Becchetti et al., SODA 2017) suggests that observing the temporal evolution of the power method applied to an initial random vector may, at least in some cases, provide enough information on the space spanned by the first two eigenvectors, so as to allow recovery of a hidden partition without explicit eigenvector computations. While the results of Becchetti et al. apply to perfectly balanced partitions and/or graphs that exhibit very strong forms of regularity, we extend their approach to graphs containing a hidden
partition and characterized by a milder form of volume-regularity. We show that the class of -volume regular graphs is the largest class of undirected (possibly weighted) graphs whose transition matrix admits stepwise eigenvectors (i.e., vectors that are constant over each set of the hidden partition). To obtain this result, we highlight a connection between volume regularity and lumpability of Markov chains. Moreover, we prove that if the stepwise eigenvectors are those associated to the first eigenvalues and the gap between the -th and the ( +1)-th eigenvalues is sufficiently large, the Averaging dynamics of Becchetti et al. recovers the underlying community structure of the graph in logarithmic time, with high probability.
Steven Kelk, Simone Linz
2018-11-16
In 2001 Allen and Steel showed that, if subtree and chain reduction rules have been applied to two unrooted phylogenetic trees, the reduced trees will have at most 28k taxa where k is the TBR (Tree Bisection and Reconnection) distance between the two trees. Here we reanalyse Allen and Steel's kernelization algorithm and prove that the reduced instances will in fact have at most 15k-9 taxa. Moreover we show, by describing a family of instances which have exactly 15k-9 taxa after reduction, that this new bound is tight. These instances also have no common clusters, showing that a third commonly-encountered reduction rule, the cluster reduction, cannot further reduce the size of the kernel in the worst case. To achieve these results we introduce and use "unrooted generators" which are analogues of rooted structures that have appeared earlier in the phylogenetic networks literature. Using similar argumentation we show that, for the minimum hybridization problem on two rooted trees, 9k-2 is a tight bound (when subtree and chain reduction rules have been applied) and 9k-4 is a tight bound (when, additionally, the cluster reduction has been applied) on the number of taxa, where k is the hybridization number of the two trees.
Chiranjib Bhattacharyya, Ravindran Kannan
2019-04-14
In this paper we show that a large class of Latent variable models, such as Mixed Membership Stochastic Block(MMSB) Models, Topic Models, and Adversarial Clustering, can be unified through a geometric perspective, replacing model specific assumptions and algorithms for individual models. The geometric perspective leads to the formulation: \emph{find a latent
polytope in given data points, each obtained by perturbing a latent point in }. This problem does not seem to have been considered in the literature. The most important contribution of this paper is to show that the latent polytope problem admits an efficient algorithm under deterministic assumptions which naturally hold in Latent variable models considered in this paper. ur algorithm runs in time matching the best running time of algorithms in special cases considered here and is better when the data is sparse, as is the case in applications. An important novelty of the algorithm is the introduction of \emph{subset smoothed polytope}, , the convex hull of the points obtained by averaging all subsets of the data points, for a given . We show that and are close in Hausdroff distance. Among the consequences of our algorithm are the following: (a) MMSB Models and Topic Models: the first quasi-input-sparsity time algorithm for parameter estimation for , (b) Adversarial Clustering: In means, if, an adversary is allowed to move many data points from each cluster an arbitrary amount towards the convex hull of the centers of other clusters, our algorithm still estimates cluster centers well.
Walter Hussak, Amitabh Trehan
2019-07-16
Flooding is among the simplest and most fundamental of all distributed network algorithms. A node begins the process by sending a message to all its neighbours and the neighbours, in the next round forward the message to all the neighbours they did not receive the message from and so on. We assume that the nodes do not keep a record of the flooding event. We call this amnesiac flooding (AF). Since the node forgets, if the message is received again in subsequent rounds, it will be forwarded again raising the possibility that the message may be circulated infinitely even on a finite graph. As far as we know, the question of termination for such a flooding process has not been settled - rather, non-termination is implicitly assumed. In this paper, we show that synchronous AF always terminates on any arbitrary finite graph and derive exact termination times which differ sharply in bipartite and non-bipartite graphs. Let
be a finite connected graph. We show that synchronous AF from a single source node terminates on in rounds, where is the eccentricity of the source node, if and only if is bipartite. For non-bipartite , synchronous AF from a single source terminates in rounds where and is the diameter of . This limits termination time to at most and at most for bipartite and non-bipartite graphs respectively. If communication/broadcast to all nodes is the motivation, our results show that AF is asymptotically time optimal and obviates the need for construction and maintenance of spanning structures like spanning trees. The clear separation in the termination times of bipartite and non-bipartite graphs also suggests mechanisms for distributed discovery of the topology/distances in arbitrary graphs. For comparison, we show that, in asynchronous networks, an adaptive adversary can force AF to be non-terminating.
Aditya Potukuchi
2019-07-07
Let
be a -regular hypergraph on vertices and edges. Let be the incidence matrix of and let us denote . We show that the discrepancy of is . As a corollary, this gives us that for every , the discrepancy of a random -regular hypergraph with vertices and edges is almost surely as grows. The proof also gives a polynomial time algorithm that takes a hypergraph as input and outputs a coloring with the above guarantee.
Michael Elkin, Ofer Neiman
2019-07-16
Given metric spaces
and and an ordering of , an embedding is said to have a prioritized distortion , if for any pair of distinct points in , the distortion provided by for this pair is at most . If is a normed space, the embedding is said to have prioritized dimension , if may have nonzero entries only in its first coordinates. The notion of prioritized embedding was introduced by \cite{EFN15}, where a general methodology for constructing such embeddings was developed. Though this methodology enables \cite{EFN15} to come up with many prioritized embeddings, it typically incurs some loss in the distortion. This loss is problematic for isometric embeddings. It is also troublesome for Matousek's embedding of general metrics into , which for a parameter , provides distortion and dimension . In this paper we devise two lossless prioritized embeddings. The first one is an isometric prioritized embedding of tree metrics into with dimension . The second one is a prioritized Matousek's embedding of general metrics into , which provides prioritized distortion and dimension , again matching the worst-case guarantee in the distortion of the classical Matousek's embedding. We also provide a dimension-prioritized variant of Matousek's embedding. Finally, we devise prioritized embeddings of general metrics into (single) ultra-metric and of general graphs into (single) spanning tree with asymptotically optimal distortion.
Huib Donkers, Bart M. P. Jansen
2019-06-13
For a fixed finite family of graphs
, the -Minor-Free Deletion problem takes as input a graph and an integer and asks whether there exists a set of size at most such that is -minor-free. For and this encodes Vertex Cover and Feedback Vertex Set respectively. When parameterized by the feedback vertex number of these two problems are known to admit a polynomial kernelization. Such a polynomial kernelization also exists for any containing a planar graph but no forests. In this paper we show that -Minor-Free Deletion parameterized by the feedback vertex number is MK[2]-hard for . This rules out the existence of a polynomial kernel assuming , and also gives evidence that the problem does not admit a polynomial Turing kernel. Our hardness result generalizes to any not containing a -subgraph-free graph, using as parameter the vertex-deletion distance to treewidth , where denotes the minimum treewidth of the graphs in . For the other case, where contains a -subgraph-free graph, we present a polynomial Turing kernelization. Our results extend to -Subgraph-Free Deletion.
Arnold Filtser, Lee-Ad Gottlieb, Robert Krauthgamer
2019-07-16
We investigate for which metric spaces the performance of distance labeling and of
-embeddings differ, and how significant can this difference be. Recall that a distance labeling is a distributed representation of distances in a metric space , where each point is assigned a succinct label, such that the distance between any two points can be approximated given only their labels. A highly structured special case is an embedding into , where each point is assigned a vector such that is approximately . The performance of a distance labeling or an -embedding is measured via its distortion and its label-size/dimension. We also study the analogous question for the prioritized versions of these two measures. Here, a priority order of the point set is given, and higher-priority points should have shorter labels. Formally, a distance labeling has prioritized label-size if every has label size at most . Similarly, an embedding has prioritized dimension if is non-zero only in the first coordinates. In addition, we compare these their prioritized measures to their classical (worst-case) versions. We answer these questions in several scenarios, uncovering a surprisingly diverse range of behaviors. First, in some cases labelings and embeddings have very similar worst-case performance, but in other cases there is a huge disparity. However in the prioritized setting, we most often find a strict separation between the performance of labelings and embeddings. And finally, when comparing the classical and prioritized settings, we find that the worst-case bound for label size often ``translates'' to a prioritized one, but also a surprising exception to this rule.
