Results 1  10
of
18
Selfimproving algorithms
 in SODA ’06: Proceedings of the seventeenth annual ACMSIAM symposium on Discrete algorithm
"... We investigate ways in which an algorithm can improve its expected performance by finetuning itself automatically with respect to an arbitrary, unknown input distribution. We give such selfimproving algorithms for sorting and computing Delaunay triangulations. The highlights of this work: (i) an al ..."
Abstract

Cited by 24 (4 self)
 Add to MetaCart
We investigate ways in which an algorithm can improve its expected performance by finetuning itself automatically with respect to an arbitrary, unknown input distribution. We give such selfimproving algorithms for sorting and computing Delaunay triangulations. The highlights of this work: (i) an algorithm to sort a list of numbers with optimal expected limiting complexity; and (ii) an algorithm to compute the Delaunay triangulation of a set of points with optimal expected limiting complexity. In both cases, the algorithm begins with a training phase during which it adjusts itself to the input distribution, followed by a stationary regime in which the algorithm settles to its optimized incarnation. 1
SelfAdjusting Trees in Practice for Large Text Collections
 Software  Practice and Experience
, 2002
"... Splay and randomised search trees are selfbalancing binary tree structures with little or no space overhead compared to a standard binary search tree. Both trees are intended for use in applications where node accesses are skewed, for example in gathering the distinct words in a large text collecti ..."
Abstract

Cited by 13 (4 self)
 Add to MetaCart
Splay and randomised search trees are selfbalancing binary tree structures with little or no space overhead compared to a standard binary search tree. Both trees are intended for use in applications where node accesses are skewed, for example in gathering the distinct words in a large text collection for index construction. We investigate the efficiency of these trees for such vocabulary accumulation. Surprisingly, unmodified splaying and randomised search trees are on average around 25% slower than using a standard binary tree. We investigate heuristics to limit splay tree reorganisation costs and show their effectiveness in practice. In particular, a periodic rotation scheme improves the speed of splaying by 27%, while other proposed heuristics are less effective. We also report the performance of efficient bitwise hashing and redblack trees for comparison.
Competitive Online Scheduling with Level of Service
 In Proc. 7th Annual International Computing and Combinatorics Conference
, 2000
"... Motivated by an application in thinwire visualization, we study an abstract online scheduling problem where the size of each requested service can be scaled down by the scheduler. Thus our problem embodies a notion of "Level of Service" that is increasingly important in multimedia appl ..."
Abstract

Cited by 11 (3 self)
 Add to MetaCart
Motivated by an application in thinwire visualization, we study an abstract online scheduling problem where the size of each requested service can be scaled down by the scheduler. Thus our problem embodies a notion of "Level of Service" that is increasingly important in multimedia applications. We give two schedulers FirstFit and EndFit based on two simple heuristics, and generalize them into a class of greedy schedulers. We show that both FirstFit and EndFit are 2competitive, and any greedy scheduler is 3competitive. These bounds are shown to be tight.
List update with locality of reference
 In Proceedings of the 8th Latin American Theoretical Informatics Symposium
, 2008
"... Abstract. It is known that in practice, request sequences for the list update problem exhibit a certain degree of locality of reference. Motivated by this observation we apply the locality of reference model for the paging problem due to Albers et al. [STOC 2002/JCSS 2005] in conjunction with biject ..."
Abstract

Cited by 4 (3 self)
 Add to MetaCart
Abstract. It is known that in practice, request sequences for the list update problem exhibit a certain degree of locality of reference. Motivated by this observation we apply the locality of reference model for the paging problem due to Albers et al. [STOC 2002/JCSS 2005] in conjunction with bijective analysis [SODA 2007] to list update. Using this framework, we prove that MovetoFront (MTF) is the unique optimal algorithm for list update. This addresses the open question of defining an appropriate model for capturing locality of reference in the context of list update [Hester and Hirschberg ACM Comp. Surv. 1985]. Our results hold both for the standard cost function of Sleator and Tarjan [CACM 1985] and the improved cost function proposed independently by Martínez and Roura [TCS 2000] and Munro [ESA 2000]. This result resolves an open problem of Martínez and Roura, namely proposing a measure which can successfully separate MTF from all other listupdate algorithms. 1
Offline list update is NPhard
 IN PROCEEDINGS OF THE 8TH ANNUAL EUROPEAN SYMPOSIUM (ESA 2000), VOLUME 1879 OF LNCS
, 2000
"... In the offline list update problem, we maintain an unsorted linear list used as a dictionary. Accessing the item at position i in the list costs i units. In order to reduce access cost, we are allowed to update the list at any time by transposing consecutive items at a cost of one unit. Given a seq ..."
Abstract

Cited by 4 (0 self)
 Add to MetaCart
In the offline list update problem, we maintain an unsorted linear list used as a dictionary. Accessing the item at position i in the list costs i units. In order to reduce access cost, we are allowed to update the list at any time by transposing consecutive items at a cost of one unit. Given a sequence of requests one has to serve in turn, we are interested in the minimal cost needed to serve all requests. Little is known about this problem. The best algorithm so far needs exponential time in the number of items in the list. We show that there is no polynomial algorithm unless P = NP.
Learning when concepts abound
, 2006
"... Many learning tasks, such as largescale text categorization and word prediction, can benefit from efficient training and classification when the number of classes, in addition to instances and features, is large, that is, in the thousands and beyond. We investigate the learning of sparse class indi ..."
Abstract

Cited by 4 (2 self)
 Add to MetaCart
Many learning tasks, such as largescale text categorization and word prediction, can benefit from efficient training and classification when the number of classes, in addition to instances and features, is large, that is, in the thousands and beyond. We investigate the learning of sparse class indices to address this challenge. An index is a mapping from features to classes. We compare the indexlearning methods against other techniques, including oneversusrest and topdown classification using perceptrons and support vector machines. We find that index learning is highly advantageous for space and time efficiency, at both training and classification times. Moreover, this approach yields similar and at times better accuracies. On problems with hundreds of thousands of instances and thousands of classes, the index is learned in minutes, while other methods can take hours or days. As we explain, the design of the learning update enables conveniently constraining each feature to connect to a small subset of the classes in the index. This constraint is crucial for scalability. Given an instance with l active (positivevalued) features, each feature on average connecting to d classes in the index (in the order of 10s in our experiments), update and classification take O(dl log(dl)).
Dynamic LengthRestricted Coding
, 2003
"... Suppose that $S$ is a string of length $m$ drawn from an alphabet of $n$ characters, $d$ of which occur in $S$. Let $P$ be the relative frequency distribution of characters in $S$. We present a new algorithm for dynamic coding that uses at most \(\lceil \lg n \rceil 1\) bits to encode each character ..."
Abstract

Cited by 3 (2 self)
 Add to MetaCart
Suppose that $S$ is a string of length $m$ drawn from an alphabet of $n$ characters, $d$ of which occur in $S$. Let $P$ be the relative frequency distribution of characters in $S$. We present a new algorithm for dynamic coding that uses at most \(\lceil \lg n \rceil 1\) bits to encode each character in $S$
A New Lower Bound for the List Update Problem in the Partial Cost Model
, 1999
"... The optimal competitive ratio for a randomized online list update algorithm is known to be at least 1.5 and at most 1.6, but the remaining gap is not yet closed. We present a new lower bound of 1.50084 for the partial cost model. The construction is based on game trees with incomplete information, w ..."
Abstract

Cited by 3 (2 self)
 Add to MetaCart
The optimal competitive ratio for a randomized online list update algorithm is known to be at least 1.5 and at most 1.6, but the remaining gap is not yet closed. We present a new lower bound of 1.50084 for the partial cost model. The construction is based on game trees with incomplete information, which seem to be generally useful for the competitive analysis of online algorithms.
Ranked Recall: Efficient classification by efficient learning of indices that rank
, 2007
"... Efficient learning and categorization in the face of myriad categories and instances is an important challenge. We investigate algorithms that efficiently learn sparse but accurate category indices. An index is a weighted bipartite graphs mapping features to categories. Given an instance, the index ..."
Abstract

Cited by 3 (3 self)
 Add to MetaCart
Efficient learning and categorization in the face of myriad categories and instances is an important challenge. We investigate algorithms that efficiently learn sparse but accurate category indices. An index is a weighted bipartite graphs mapping features to categories. Given an instance, the index retrieves, scores, and ranks a set of candidate categories. The ranking or the scores can then be used for category assignment. We compare index learning against other classification approaches, including oneversusrest and topdown classification using support vector machines. We find that the indexing approach is highly advantageous in terms of space and time efficiency, at both training and classification times, while retaining competitive accuracy. On problems with hundreds of thousands of instances and thousands of categories, the index is learned in minutes, while other methods can take orders of magnitude longer.
List Update with Locality of Reference: MTF Outperforms All Other Algorithms
"... It has been observed that in practice, typical request sequences for the list update problem exhibit a certain degree of locality of reference. We first extend the locality of reference model for the paging problem due to Albers et al [STOC 2002/JCSS 2005] to the domain of list update; this addresse ..."
Abstract

Cited by 2 (0 self)
 Add to MetaCart
It has been observed that in practice, typical request sequences for the list update problem exhibit a certain degree of locality of reference. We first extend the locality of reference model for the paging problem due to Albers et al [STOC 2002/JCSS 2005] to the domain of list update; this addresses the open question of defining an appropriate model for capturing locality of reference in the context of list update [Hester and Hirschberg ACM Comp. Surv. 1985]. We then apply this model in conjunction with a recent technique for comparing online algorithms, namely bijective analysis [SODA 2007] and analyze well known online algorithms for list update. Using this framework, we prove that MovetoFront (MTF) is the unique optimal algorithm for list update. This holds for both the standard cost function of Sleator and Tarjan [C. ACM 1985] and the refined cost function proposed independently by Martínez and Roura [TCS 2000] and Munro [ESA 2000]. Our work resolves an open conjecture of Martínez and Roura, namely proposing a measure which can successfully separate MTF from all other algorithms. 1