A common way to model multi-class classification problems is by means of Error-Correcting Output Codes (ECOC). One of the main requirements of the ECOC design is that the base classifier is capable of splitting each sub-group of classes from each binary problem. In this paper, we present a novel strategy to model multi-class classification problems using sub-class information in the ECOC framew...

In this paper, we consider two restricted types of dynamic orthogonal segment intersection search problems. One is the problem in which the underlying set is updated only by insertions. In the other, the set is updated only by deletions. We show that an intermixed sequence of O ( n ) queries and updates in both problems can be executed on-line in O(n1ognSK) time and O(n1ogn) space, where K is t...

We present a framework for unifying iteration reordering transformations such as loop interchange, loop distribution, skewing, tiling, index set splitting and statement reordering. The framework is based on the idea that a transformation can be represented as a mapping from the original iteration space to a new iteration space. The framework is designed to provide a uniform way to represent and...

This paper presents a new unified method for simultaneously tiling the register and cache levels of the memory hierarchy. We will focus on the code transformation phase of tiling. Our algorithm uses strip-mining and loop interchange on all memory hierarchy levels to determine the tiles as usual, and, afterwards, and due to the special characteristics of the register level, we apply index set sp...

In this paper we present an extensive study of weight pruning and regularization techniques for feed forward neural nets. This algorithm comparison is based on a data base of known benchmark data sets and a realistic real world data set from automotive control. We trained 30 nets for every algorithm implemented by us, with three different ways of data set splitting. For the 38 implemented metho...

The two objectives of this paper are: (1) to articulate three new general techniques for designing FPT algorithms, and (2) to apply these to obtain new FPT algorithms for Set Splitting and Vertex Cover. In the case of Set Splitting, we improve the best previous O(72) FPT algorithm due to Dehne, Fellows and Rosamond [DFR03], to O(8) by an approach based on greedy localization in conjunction with...

The Birkhoff-von Neumann (BVN) strategy for offline switching does not support multicast, as it considers only permutation-based switch configurations. This thesis extends the BVN strategy to multicast switching. Using a graph theoretic model, we show that the capacity region for a traffic pattern is precisely the stable set polytope of the pattern's "conflict graph", in the no-fanout-splitting...

The Birkhoff-von Neumann (BVN) strategy for offline switching does not support multicast, as it considers only permutation-based switch configurations. This thesis extends the BVN strategy to multicast switching. Using a graph theoretic model, we show that the capacity region for a traffic pattern is precisely the stable set polytope of the pattern’s “conflict graph”, in the no-fanout-splitting...

The theory of regions was introduced by Ehrenfeucht and Rozenberg in the early nineties to explain how to derive (synthesize) an event-based model from an automaton. To be applicable, the theory relies on stringent conditions on the input automaton. Although some relaxation on these restrictions has been done in the last decade, in general not every automaton can be synthesized while preserving...

We study the problem k-Set Splitting in fixed parameter complexity. We show that the problem can be solved in time O∗(2.6494k), improving on the best currently known running time of O∗(8k). This is done by showing that a non-trivial instance must have a small minimal Set Cover, and using this to reduce the problem to a series of small instances of Max Sat. We also give a linear kernel containin...