TensorFlow is a machine learning system that operates at large scale and in heterogeneous environments. Tensor-Flow uses dataflow graphs to represent computation, shared state, and the operations that mutate that state. It maps the nodes of a dataflow graph across many machines in a cluster, and within a machine across multiple computational devices, including multicore CPUs, general-purpose GPUs, and custom-designed ASICs known as Tensor Processing Units (TPUs). This architecture gives flexibility to the application developer: whereas in previous "parameter server" designs the management of shared state is built into the system, TensorFlow enables developers to experiment with novel optimizations and training algorithms. TensorFlow supports a variety of applications, with a focus on training and inference on deep neural networks. Several Google services use TensorFlow in production, we have released it as an open-source project, and it has become widely used for machine learning research. In this paper, we describe the TensorFlow dataflow model and demonstrate the compelling performance that TensorFlow achieves for several real-world applications.

/pdf/tensorflow-a-system-for-large-scale-machine-learning-10wgdjhl6k.pdf

TensorFlow: a system for large-scale machine learning

Anomaly detection is an important problem that has been researched within diverse research areas and application domains. Many anomaly detection techniques have been specifically developed for certain application domains, while others are more generic. This survey tries to provide a structured and comprehensive overview of the research on anomaly detection. We have grouped existing techniques into different categories based on the underlying approach adopted by each technique. For each category we have identified key assumptions, which are used by the techniques to differentiate between normal and anomalous behavior. When applying a given technique to a particular domain, these assumptions can be used as guidelines to assess the effectiveness of the technique in that domain. For each category, we provide a basic anomaly detection technique, and then show how the different existing techniques in that category are variants of the basic technique. This template provides an easier and more succinct understanding of the techniques belonging to each category. Further, for each category, we identify the advantages and disadvantages of the techniques in that category. We also provide a discussion on the computational complexity of the techniques since it is an important issue in real application domains. We hope that this survey will provide a better understanding of the different directions in which research has been done on this topic, and how techniques developed in one area can be applied in domains for which they were not intended to begin with.

/pdf/anomaly-detection-a-survey-1x33g9m0a3.pdf

Anomaly detection: A survey

TensorFlow is a machine learning system that operates at large scale and in heterogeneous environments. TensorFlow uses dataflow graphs to represent computation, shared state, and the operations that mutate that state. It maps the nodes of a dataflow graph across many machines in a cluster, and within a machine across multiple computational devices, including multicore CPUs, general-purpose GPUs, and custom designed ASICs known as Tensor Processing Units (TPUs). This architecture gives flexibility to the application developer: whereas in previous "parameter server" designs the management of shared state is built into the system, TensorFlow enables developers to experiment with novel optimizations and training algorithms. TensorFlow supports a variety of applications, with particularly strong support for training and inference on deep neural networks. Several Google services use TensorFlow in production, we have released it as an open-source project, and it has become widely used for machine learning research. In this paper, we describe the TensorFlow dataflow model in contrast to existing systems, and demonstrate the compelling performance that TensorFlow achieves for several real-world applications.

TensorFlow: A system for large-scale machine learning

XML is quickly becoming the de facto standard for data exchange over the Internet. This is creating a new set of data management requirements involving XML, such as the need to store and query XML documents. Researchers have proposed using relational database systems to satisfy these requirements by devising ways to "shred" XML documents into relations, and translate XML queries into SQL queries over these relations. However, a key issue with such an approach, which has largely been ignored in the research literature, is how (and whether) the ordered XML data model can be efficiently supported by the unordered relational data model. This paper shows that XML's ordered data model can indeed be efficiently supported by a relational database system. This is accomplished by encoding order as a data value. We propose three order encoding methods that can be used to represent XML order in the relational data model, and also propose algorithms for translating ordered XPath expressions into SQL using these encoding methods. Finally, we report the results of an experimental study that investigates the performance of the proposed order encoding methods on a workload of ordered XML queries and updates.

http://db.ucsd.edu/static/cse232b-archive/papers/tatarinov02.pdf

Storing and querying ordered XML using a relational database system

In this paper, we review the background and state-of-the-art of big data. We first introduce the general background of big data and review related technologies, such as could computing, Internet of Things, data centers, and Hadoop. We then focus on the four phases of the value chain of big data, i.e., data generation, data acquisition, data storage, and data analysis. For each phase, we introduce the general background, discuss the technical challenges, and review the latest advances. We finally examine the several representative applications of big data, including enterprise management, Internet of Things, online social networks, medial applications, collective intelligence, and smart grid. These discussions aim to provide a comprehensive overview and big-picture to readers of this exciting area. This survey is concluded with a discussion of open problems and future directions.

Big Data: A Survey

Disclosed is a technique for indexing data. For each token in a set of documents, a sort key is generated that includes a document identifier that indicates whether a section of a document associated with the sort key is an anchor text section or a context section, wherein the anchor text section and the context text section have a same document identifier; it is determined whether a data field associated with the token is a fixed width; when the data field is a fixed width, the token is designated as one for which fixed width sort is to be performed; and, when the data field is a variable length, the token is designated as one for which a variable width sort is to be performed. The fixed width sort and the variable width sort are performed. For each document, the sort keys are used to bring together the anchor text section and the context section of that document.

Architecture for an indexer

This paper describes Cricket, a new database storage system that is intended to be used as a platform for design environments and persistent programming languages Cricket uses the memory management primitives of the Mach operating system to provide the abstraction of a shared, transactional single-level store that can be directly accessed by user applications In this paper, we present the design and motivation for Cricket We also present some initial performance results which show that, for its intended applications, Cricket can provide better performance than a general-purpose database storage system

/pdf/cricket-a-mapped-persistent-object-store-37o2j1pxay.pdf

Cricket: A Mapped, Persistent Object Store

A procedure for detecting a reordering requirement in a directed record stream during query execution in a relational database processing system. The query compiler component of a relational database processing system includes procedures for building query execution plans (QEPs) for evaluation preparatory to selecting an optimal plan for execution. These plans are constructed from the bottom up using an internal graphical representation for the user query that has a number of relation nodes interconnected by directed record streams (data flows). A relational operation within each node imposes an "order requirement" on the outflow stream represented by an order requirement vector OR. The records within each directed record stream have an "order property" represented by an order property vector OP. Order detection occurs when these two vectors are compared to determine whether the order property satisfies the order requirement. Order detection by normalization (ODN) according to this invention first normalizes the two order specification vectors to remove all attributes made redundant by the effects of predicates and functional dependencies. Query execution plans constructed using ODN are found to execute an order of magnitude faster than those constructed using order detection without normalization.

Semantic optimization of query order requirements using order detection by normalization in a query compiler system

Unwieldy star/join queries are performed more efficiently using a filtered fact table. Suitable queries include star/join queries with a large fact table joined with multiple subsidiary dimension tables, where indices exist over fact table join columns. The query is analyzed to prepare a query plan for the dimension table accesses. This plan is supplemented by adding nested loop join operations, where the inner table is a dimension table plan and the outer table is an index scan performed over a fact table index of the join column with the dimension table. The plan is also supplemented by filtering records resulting from the nested loop joins using a sequence of dynamic bit vectors, ultimately yielding a list of probable fact table records. The plan is further supplemented by fetching these records to construct a distilled fact which is used, instead of the large original table, to execute the query in considerably less time. If desired, the supplemented query plan and other competing approaches may studied to provide cost estimates, with the least costly approach being actually implemented.

Star/join query optimization

A technique enabling querying of XML documents in a relational database system via a reconstruction view allowing XML documents to be queried as though XML views of relational data. A single query processor can be used with all relational schema generation methods (whether or not using XML schema information) to seamlessly query across XML documents, relational data, and XML views of relational data; no special purpose query processor is required. The technique creates an XML document view, creates relational tables for storing XML documents using relational schema, shreds the XML documents and stores the XML documents as rows in the relational tables according to the relational schema, generates a reconstruction view over the relational tables to define how the shredded documents are to be virtually reconstructed, and processes queries over the stored XML documents as queries over the reconstruction view.

Eugene J. Shekita

Papers

Architecture for an indexer

Cricket: A Mapped, Persistent Object Store

Semantic optimization of query order requirements using order detection by normalization in a query compiler system

Star/join query optimization

System, method, and computer program product for querying XML documents using a relational database system