Investigation of acoustic units for LVCSR systems

One important issue in designing state-of-the-art LVCSR systems is the choice of acoustic units. Context dependent (CD) phones remain the dominant form of acoustic units. They can capture the co-articulatory effect in speech via explicit modelling. However, for other more complicated phonological processes, they rely on the implicit modelling ability of the underlying statistical models. Alternatively, it is possible to construct acoustic models based on higher level linguistic units, for example, syllables, to explicitly capture these complex patterns. When sufficient training data is available, this approach may show an advantage over implicit acoustic modelling. In this paper a wide range of acoustic units are investigated to improve LVCSR system performance. Significant error rate gains up to 7.1% relative (0.8% abs.) were obtained on a state-of-the-art Mandarin Chinese broadcast audio recognition task using word and syllable position dependent triphone and quinphone models. © 2011 IEEE.

BarraCUDA – a Fast Sequence Mapping Software using Graphics Processing Units

High-throughput DNA sequencing (HTS) instruments today are capable of generating millions of sequencing reads in a short period of time, and this represents a serious challenge to current bioinformatics pipeline in processing such an enormous amount of data in a fast and economical fashion. Modern graphics cards are powerful processing units that consist of hundreds of scalar processors in parallel in order to handle the rendering of high-definition graphics in real-time. It is this computational capability that we propose to harness in order to accelerate some of the time-consuming steps in analyzing data generated by the HTS instruments. We have developed BarraCUDA, a novel sequence mapping software that utilizes the parallelism of NVIDIA CUDA graphics cards to map sequencing reads to a particular location on a reference genome. While delivering a similar mapping fidelity as other mainstream programs , BarraCUDA is a magnitude faster in mapping throughput compared to its CPU counterparts. The software is also capable of supporting multiple CUDA devices in parallel to further accelerate the mapping throughput. BarraCUDA is designed to take advantage of the parallelism of GPU to accelerate the mapping of millions of sequencing reads generated by HTS instruments. By doing this, we could, at least in part streamline the current bioinformatics pipeline such that the wider scientific community could benefit from the sequencing technology. BarraCUDA is currently available at http://seqbarracuda.sf.net

BarraCUDA - a fast short read sequence aligner using graphics processing units.

BACKGROUND: With the maturation of next-generation DNA sequencing (NGS) technologies, the throughput of DNA sequencing reads has soared to over 600 gigabases from a single instrument run. General purpose computing on graphics processing units (GPGPU), extracts the computing power from hundreds of parallel stream processors within graphics processing cores and provides a cost-effective and energy efficient alternative to traditional high-performance computing (HPC) clusters. In this article, we describe the implementation of BarraCUDA, a GPGPU sequence alignment software that is based on BWA, to accelerate the alignment of sequencing reads generated by these instruments to a reference DNA sequence. FINDINGS: Using the NVIDIA Compute Unified Device Architecture (CUDA) software development environment, we ported the most computational-intensive alignment component of BWA to GPU to take advantage of the massive parallelism. As a result, BarraCUDA offers a magnitude of performance boost in alignment throughput when compared to a CPU core while delivering the same level of alignment fidelity. The software is also capable of supporting multiple CUDA devices in parallel to further accelerate the alignment throughput. CONCLUSIONS: BarraCUDA is designed to take advantage of the parallelism of GPU to accelerate the alignment of millions of sequencing reads generated by NGS instruments. By doing this, we could, at least in part streamline the current bioinformatics pipeline such that the wider scientific community could benefit from the sequencing technology.BarraCUDA is currently available from http://seqbarracuda.sf.net.

A cost-effective atomic force microscope for undergraduate control laboratories

This paper presents a simple, cost-effective and robust atomic force microscope (AFM), which has been purposely designed and built for use as a teaching aid in undergraduate controls labs. The guiding design principle is to have all components be open and visible to the students, so the inner functioning of the microscope has been made clear to see. All of the parts but one are off the shelf, and assembly time is generally less than two days, which makes the microscope a robust instrument that is readily handled by the students with little chance of damage. While the scanning resolution is nowhere near that of a commercial instrument, it is more than sufficient to take interesting scans of micrometer-scale objects. A survey of students after their having used the AFM resulted in a generally good response, with 80% agreeing that they had a positive learning experience. © 2009 IEEE.

Shifting atomic patterns: on the origin of the different atomic-scale patterns of graphite as observed using scanning tunnelling microscopy.

We present an in-depth study of the myriad atomically resolved patterns observed on graphite using the scanning tunnelling microscope (STM) over the past three decades. Through the use of highly resolved atomic resolution images, we demonstrate how the interactions between the different graphene layers comprising graphite affect the local surface atomic charge density and its resulting symmetry orientation, with particular emphasis on interactions that are thermodynamically unstable. Moreover, the interlayer graphene coupling is controlled experimentally by varying the tip-surface interaction, leading to associated changes in the atomic patterns. The images are corroborated by first-principles calculations, further validating our claim that surface graphene displacement, coming both from lateral and vertical displacement of the top graphene layer, forms the basis of the rich variety of atomic patterns observed in STM experiments on graphite.

Substrate-assisted nucleation of ultra-thin dielectric layers on graphene by atomic layer deposition

We report on a large improvement in the wetting of Al 2O 3 thin films grown by un-seeded atomic layer deposition on monolayer graphene, without creating point defects. This enhanced wetting is achieved by greatly increasing the nucleation density through the use of polar traps induced on the graphene surface by an underlying metallic substrate. The resulting Al 2O 3/graphene stack is then transferred to SiO 2 by standard methods. © 2012 American Institute of Physics.