This paper introduces a modified shifting bottleneck approach to solve train scheduling and rescheduling problems. The problem is formulated as a job shop scheduling model and a mixed integer linear programming model is also presented. The shifting bottleneck procedure is a wellestablished heuristic method for obtaining solutions to the job shop and other machine scheduling problems. We modify ...

This paper reports the state of development of PThomas, a network based document retrieval system implemented on a massively parallel fine-grained computer, the Connection Machine. The program is written in C*, an enhancement of the C programming language which exploits the parallelism of the Connection Machine. The system is based on Oddy’s original Thomas program, which was highly parallel in...

We usually think of an audio application as a self-contained executable that will compute audio, allow user interface control, and render sound in a single process, on a unique machine. With the appearance of fast network and sophisticated, light and wireless control devices (such as tablets, smartphones...) the three different parts (that are audio computation, interface control and sound rend...

In this brief report we compared the performance of selected scientific applications on sun-test a massive SMP 32-core machine with blade and zebra, the two main partitions of the production cluster at SISSA. In tables 1 and 2 some details on the hardware are presented. Information on the interconnection network is available in table 3 along with some network performance results, obtained with ...

We describe the neural-network training framework used in the Kaldi speech recognition toolkit, which is geared towards training DNNs with large amounts of training data using multiple GPU-equipped or multi-core machines. In order to be as hardware-agnostic as possible, we needed a way to use multiple machines without generating excessive network traffic. Our method is to average the neural net...

We describe the neural-network training framework used in the Kaldi speech recognition toolkit, which is geared towards training DNNs with large amounts of training data using multiple GPU-equipped or multicore machines. In order to be as hardwareagnostic as possible, we needed a way to use multiple machines without generating excessive network traffic. Our method is to average the neural netwo...

This paper presents a neural network solution that combines character segmentation and character recognition concurrently as a single task. Current segmentation methods utilize traditional image processing techniques such as spatial histograms which are only 60% accurate on handprint. Using traditional techniques for segmenting handprint in a model recognition system running on a massively para...

In the last decade, many systems for the extraction of operational statistics from computer network interconnects have been designed and implemented. Those systems generate huge amounts of data of various formats and in various granularities, from packet level to statistics about whole flows. In addition, the complexity of Internet services has increased drastically with the introduction of clo...

The quality of machine translation is rapidly evolving. Today one can find several machine translation systems on the web that provide reasonable translations, although the systems are not perfect. In some specific domains, the quality may decrease. A recently proposed approach to this domain is neural machine translation. It aims at building a jointly-tuned single neural network that maximizes...