Publications

The success of MapReduce has sparked the development of a diverse array of cluster computing frameworks. We believe that no single framework will be optimal for all applications, and that organizations will instead want to run multiple frameworks in the same cluster. Furthermore, to ease development of new frameworks, it is critical to identify common abstractions and modularize their architectures. To achieve these goals, we propose Nexus, a low-level substrate that provides isolation and efficient resource sharing across frameworks running on the same cluster, while giving each framework maximum control over the scheduling and execution of its jobs. Nexus fosters innovation in the cloud by letting organizations run new frameworks alongside existing ones and by letting framework developers focus on specific applications rather than building one-size-fits-all frameworks.

Scalable distributed dataflow systems have recently experienced widespread adoption, with commodity dataflow engines such as Hadoop and Spark, and even commodity SQL engines routinely supporting increasingly sophisticated analytics tasks (e.g., support vector machines, logistic regression, collaborative filtering). However, these systems' synchronous (often Bulk Synchronous Parallel) dataflow execution model is at odds with an increasingly important trend in the machine learning community: the use of asynchrony via shared, mutable state (i.e., data races) in convex programming tasks, which has---in a single-node context---delivered noteworthy empirical performance gains and inspired new research into asynchronous algorithms. In this work, we attempt to bridge this gap by evaluating the use of lightweight, asynchronous state transfer within a commodity dataflow engine. Specifically, we investigate the use of asynchronous sideways information passing (ASIP) that presents single-stage parallel iterators with a Volcano-like intra-operator iterator that can be used for asynchronous information passing. We port two synchronous convex programming algorithms, stochastic gradient descent and the alternating direction method of multipliers (ADMM), to use ASIPs. We evaluate an implementation of ASIPs within on Apache Spark that exhibits considerable speedups as well as a rich set of performance trade-offs in the use of these asynchronous algorithms.

While there have been significant gains in the field of automated video description, the generalization performance of automated description models to novel domains remains a major barrier to using these systems in the real world. Most visual description methods are known to capture and exploit patterns in the training data leading to evaluation metric increases, but what are those patterns? In this work, we examine several popular visual description datasets, and capture, analyze, and understand the dataset-specific linguistic patterns that models exploit but do not generalize to new domains. At the token level, sample level, and dataset level, we find that caption diversity is a major driving factor behind the generation of generic and uninformative captions. We further show that state-of-the-art models even outperform held-out ground truth captions on modern metrics, and that this effect is an artifact of linguistic diversity in datasets. Understanding this linguistic diversity is key to building strong captioning models, we recommend several methods and approaches for maintaining diversity in the collection of new data, and dealing with the consequences of limited diversity when using current models and metrics.

The mixed-mode delaminating beam (MMDB) is a widely used test geometry designed to measure the fracture resistance of bimaterial interfaces under mixed-mode loading conditions. In the present work, linear-elastic finite element analyses are employed to determine the complex stress intensity factor, K, for an interfacial crack in this sample; results are found to confirm those of previous studies of the bilayer specimen. However, the numerical results further reveal that the region of K-dominance near the crack tip is very limited, extending merely ∼ 1/100 to 1/1000 of the sample height, about an order of magnitude smaller than for other common fracture-mechanics test samples. Analyses performed for this specimen geometry modified to include a thin sandwiched interlayer also indicate a very limited region of K-dominance, for example, extending ∼ 1/10 of the height of the sandwiched layer from the crack tip for very thin sandwiched layers. For the sandwiched geometry, two situations leading to a useful small scale yielding condition are described. Examples of the use of this specimen in the evaluation of the fracture toughness of ceramic joints are cited from the literature and comparisons are made between the size of the K-dominant region and the extent of crack tip plasticity. Based on these comparisons, the geometry-independent predictive power of linear-elastic fracture mechanics for common bimaterial couples and specimen dimensions using this geometry is challenged.

In 1971, Dr. Leon Chua, known as the father of nonlinear circuits and cellular neural networks, postulated the existence of memristor, a portmanteau of memory resistor, in his seminal paper: Memristor-the missing circuit element published in IEEE Transactions on Circuit Theory, the predecessor of IEEE Transactions on Circuits and Systems. Thirty-seven years after he predicted its existence, in the May 1 (2008) issue of the journal <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Nature</i> , a team at HP Labs led by the scientist R. S. Williams proved that the memristor was real by formulating a physics-based model of a memristor and build nanoscale devices in their lab that demonstrate all of the necessary operating characteristics. Since then, the extensive interest of academic and industrial circles on neuromorphic computing based on memristor has been skyrocketed. Moreover, the unusual electrical properties of circuits and systems based on memristor can mimic the functionalities of the human brain, and can provide an in-depth understanding of key design implications of memristor-based memories, such as learning and anticipating. As a result, neuromorphic computing based on memristor is expected to bring significant breakthrough in dynamic neuromorphic memories, memristor-based resistive RAM, non-volatile memory technology, and so on.

Search advertising shows trends of vertical extension. Vertical ads typically offer better Return of Investment (ROI) to advertisers as a result of better user engagement. However, campaign and bids in vertical ads are not set at the keyword level. As a result, the matching between user query and ads suffers low recall rate and the match quality is heavily impacted by tail queries. In this paper, we propose a retail ads retrieval framework based on query rewrite using personal history data to improve ads recall rate. To insure ads quality, we also present a relevance model for matching rewritten queries with user search intent, with a particular focus on rare queries. Extensive experiments are carried out on large-scale logs collected from the Bing search engine, and results show our system achieves significant gains in ads retrieval rate without compromising ads quality. To our knowledge, this work is the first attempt to leverage user behavioral data in ad matching and apply it to the vertical ads domain.

A functional 2-C-methyl-D-erythritol 4-phosphate (MEP) pathway is required for isoprenoid biosynthesis and hence survival in Escherichia coli and most other bacteria. In the first two steps of the pathway, MEP is produced from the central metabolic intermediates pyruvate and glyceraldehyde 3-phosphate via 1-deoxy-D-xylulose 5-phosphate (DXP) by the activity of the enzymes DXP synthase (DXS) and DXP reductoisomerase (DXR). Because the MEP pathway is absent from humans, it was proposed as a promising new target to develop new antibiotics. However, the lethal phenotype caused by the deletion of DXS or DXR was found to be suppressed with a relatively high efficiency by unidentified mutations. Here we report that several mutations in the unrelated genes aceE and ribB rescue growth of DXS-defective mutants because the encoded enzymes allowed the production of sufficient DXP in vivo. Together, this work unveils the diversity of mechanisms that can evolve in bacteria to circumvent a blockage of the first step of the MEP pathway.

Abstract not Available.