High Performance Computing (HPC) demand is on the rise, particularly for large distributed computing. HPC systems have, by design, very heterogeneous architectures, both in computation and in communication bandwidth, resulting in wide variations in the cost of communications between compute units. If large distributed applications are to take full advantage of HPC, the physical communication capabilities must be taken into consideration when allocating workload. Hypergraphs are good at modelling total volume of communication in parallel and distributed applications. To the best of our knowledge, there are no hypergraph partitioning algorithms to date that are architecture-aware. We propose a novel restreaming hypergraph partitioning algorithm (HyperPRAW) that takes advantage of peer to peer physical bandwidth profiling data to improve distributed applications performance in HPC systems. Our results show that not only the quality of the partitions achieved by our algorithm is comparable with state-of-the-art multilevel partitioning, but that the runtime performance in a synthetic benchmark is significantly reduced in 10 hypergraph models tested, with speedup factors of up to 14x.

There is extensive evidence that biological neural networks encode information in the precise timing of the spikes generated and transmitted by neurons, which offers several advantages over rate-based codes. Here we adopt a vector space formulation of spike train sequences and introduce a new liquid state machine (LSM) network architecture and a new forward orthogonal regression algorithm to learn an input-output signal mapping or to decode the brain activity. The proposed algorithm uses precise spike timing to select the presynaptic neurons relevant to each learning task. We show that using precise spike timing to train the LSM and selecting the readout presynaptic neurons leads to a significant increase in performance on binary classification tasks, in decoding neural activity from multielectrode array recordings, as well as in a speech recognition task, compared with what is achieved using the standard architecture and training methods.

Inferring mathematical models of sensory processing systems directly from input-output observations, while making the fewest assumptions about the model equations and the types of measurements available, is still a major issue in computational neuroscience. This letter introduces two new approaches for identifying sensory circuit models consisting of linear and nonlinear filters in series with spiking neuron models, based only on the sampled analog input to the filter and the recorded spike train output of the spiking neuron. For an ideal integrate-and-fire neuron model, the first algorithm can identify the spiking neuron parameters as well as the structure and parameters of an arbitrary nonlinear filter connected to it. The second algorithm can identify the parameters of the more general leaky integrate-and-fire spiking neuron model, as well as the parameters of an arbitrary linear filter connected to it. Numerical studies involving simulated and real experimental recordings are used to demonstrate the applicability and evaluate the performance of the proposed algorithms.

Integrate-and-fire neurons are time encoding machines that convert the amplitude of an analog signal into a nonuniform, strictly increasing sequence of spike times. Under certain conditions, the encoded signals can be reconstructed from the nonuniform spike time sequences using a time decoding machine. Time encoding and time decoding methods have been studied using the nonuniform sampling theory for band-limited spaces, as well as for generic shift-invariant spaces. This letter proposes a new framework for studying IF time encoding and decoding by reformulating the IF time encoding problem as a uniform sampling problem. This framework forms the basis for two new algorithms for reconstructing signals from spike time sequences. We demonstrate that the proposed reconstruction algorithms are faster, and thus better suited for real-time processing, while providing a similar level of accuracy, compared to the standard reconstruction algorithm.

MOTIVATION: Human pluripotent stem cell lines persist in culture as a heterogeneous population of SSEA3 positive and SSEA3 negative cells. Tracking individual stem cells in real time can elucidate the kinetics of cells switching between the SSEA3 positive and negative substates. However, identifying a cell's substate at all time points within a cell lineage tree is technically difficult. RESULTS: A variational Bayesian Expectation Maximization (EM) with smoothed probabilities (VBEMS) algorithm for hidden Markov trees (HMT) is proposed for incomplete tree structured data. The full posterior of the HMT parameters is determined and the underflow problems associated with previous algorithms are eliminated. Example results for the prediction of the types of cells in synthetic and real stem cell lineage trees are presented. AVAILABILITY: The Matlab code for the VBEMS algorithm is freely available at http://www.acse.dept.shef.ac.uk/repository/vbems_lineage_tree/VBEMS.ZIP CONTACT: [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: Peptide mass fingerprinting (PMF) is a method for protein identification in which a protein is fragmented by a defined cleavage protocol (usually proteolysis with trypsin), and the masses of these products constitute a 'fingerprint' that can be searched against theoretical fingerprints of all known proteins. In the first stage of PMF, the raw mass spectrometric data are processed to generate a peptide mass list. In the second stage this protein fingerprint is used to search a database of known proteins for the best protein match. Although current software solutions can typically deliver a match in a relatively short time, a system that can find a match in real time could change the way in which PMF is deployed and presented. In a paper published earlier we presented a hardware design of a raw mass spectra processor that, when implemented in Field Programmable Gate Array (FPGA) hardware, achieves almost 170-fold speed gain relative to a conventional software implementation running on a dual processor server. In this article we present a complementary hardware realization of a parallel database search engine that, when running on a Xilinx Virtex 2 FPGA at 100 MHz, delivers 1800-fold speed-up compared with an equivalent C software routine, running on a 3.06 GHz Xeon workstation. The inherent scalability of the design means that processing speed can be multiplied by deploying the design on multiple FPGAs. The database search processor and the mass spectra processor, running on a reconfigurable computing platform, provide a complete real-time PMF protein identification solution.

MOTIVATION: High-resolution mass spectrometers generate large data files that are complex, noisy and require extensive processing to extract the optimal data from raw spectra. This processing is readily achieved in software and is often embedded in manufacturers' instrument control and data processing environments. However, the speed of this data processing is such that it is usually performed off-line, post data acquisition. We have been exploring strategies that would allow real-time advanced processing of mass spectrometric data, making use of the reconfigurable computing paradigm, which exploits the flexibility and versatility of Field Programmable Gate Arrays (FPGAs). This approach has emerged as a powerful solution for speeding up time-critical algorithms. We describe here a reconfigurable computing solution for processing raw mass spectrometric data generated by MALDI-ToF instruments. The hardware-implemented algorithms for de-noising, baseline correction, peak identification and deisotoping, running on a Xilinx Virtex 2 FPGA at 180 MHz, generate a mass fingerprint over 100 times faster than an equivalent algorithm written in C, running on a Dual 3 GHz Xeon workstation.