No abstract

Deep learning has been shown as a successful machine learning method for a variety of tasks, and its popularity results in numerous open-source deep learning software tools. Training a deep network is usually a very time-consuming process. To address the computational challenge in deep learning, many tools exploit hardware features such as multi-core CPUs and many-core GPUs to shorten the training time. However, different tools exhibit different features and running performance when training different types of deep networks on different hardware platforms, which makes it difficult for end users to select an appropriate pair of software and hardware. In this paper, we aim to make a comparative study of the state-of-the-art GPU-accelerated deep learning software tools, including Caffe, CNTK, MXNet, TensorFlow, and Torch. We first benchmark the running performance of these tools with three popular types of neural networks on two CPU platforms and three GPU platforms. We then benchmark some distributed versions on multiple GPUs. Our contribution is two-fold. First, for end users of deep learning tools, our benchmarking results can serve as a guide to selecting appropriate hardware platforms and software tools. Second, for software developers of deep learning tools, our in-depth analysis points out possible future directions to further optimize the running performance.

Cognitive radio networks (CRNs) have emerged as advanced and promising paradigm to exploit the existing wireless spectrum opportunistically. It is crucial for users in CRNs to search for neighbors via rendezvous process and thereby establish the communication links to exchange the information necessary for spectrum management and channel contention etc. This paper focuses on the design of algorithms for blind rendezvous, i.e., rendezvous without using any central controller and common control channel (CCC). We propose a jump-stay based channel-hopping (CH) algorithm for blind rendezvous. The basic idea is to generate CH sequence in rounds and each round consists of a jump-pattern and a stay-pattern. Users "jump" on available channels in the jump-pattern while "stay" on a specific channel in the stay-pattern. Compared with the existing CH algorithms, our algorithm achieves the following advances: i) guaranteed rendezvous without the need of timesynchronization; ii) applicability to rendezvous of multi-user and multi-hop scenarios. We derive the maximum time-torendezvous (TTR) and the upper-bound of expected TTR of our algorithm for both 2-user and multi-user scenarios (shown in Table I). Extensive simulations are further conducted to evaluate performance of our algorithm.

SOAP3 is the first short read alignment tool that leverages the multi-processors in a graphic processing unit (GPU) to achieve a drastic improvement in speed. We adapted the compressed full-text index (BWT) used by SOAP2 in view of the advantages and disadvantages of GPU. When tested with millions of Illumina Hiseq 2000 length-100 bp reads, SOAP3 takes < 30 s to align a million read pairs onto the human reference genome and is at least 7.5 and 20 times faster than BWA and Bowtie, respectively. For aligning reads with up to four mismatches, SOAP3 aligns slightly more reads than BWA and Bowtie; this is because SOAP3, unlike BWA and Bowtie, is not heuristic-based and always reports all answers.