专利快速检索-快速检索全球专利，免费商用专利数据库-IPRDB

1. 发明申请

WO2017161275A1 TUPLE ENCODING AWARE DIRECT MEMORY ACCESS ENGINE FOR SCRATCHPAD ENABLED MULTICORE PROCESSORS 审中-公开
标题翻译： TUPLE编码唤醒直接存储器访问引擎用于支持SCRATCHPAD的多核处理器
公开(公告)号：WO2017161275A1
公开(公告)日：2017-09-21
申请号：PCT/US2017/022975
申请日：2017-03-17
申请人： ORACLE INTERNATIONAL CORPORATION
发明人： BROWN, David A. , JAIN, Rishabh , DULLER, Michael , IDICULA, Sam , SCHLANGER, Erik , HAWKINS, David Joseph
IPC分类号： G06F13/28
摘要： Tabular data is moved efficiently from a source memory to a destination memory. A first set of electronic circuits accesses a descriptor stored at a particular memory location. The descriptor indicates a width of a column and a number of rows of tabular data, and a tabular data manipulation operation, and indicates a source memory location storing the tabular data and a destination memory location for storing the tabular data manipulation result. The first set of electronic circuits determines control information indicating the tabular data manipulation operations to be performed and transmits the control information to a second set of electronic circuits. Based on the control information, the second set of electronic circuits retrieve the tabular data from the source memory location and applies the tabular data manipulation operation to generate the data manipulation result, and causes the data manipulation result to be stored at the destination memory location.
摘要翻译：
表格数据从源存储器有效地移动到目标存储器。第一组电子电路访问存储在特定存储位置的描述符。描述符指示列的宽度和表格数据的行数以及表格数据操作操作，并且指示存储表格数据的源存储器位置和用于存储表格数据操作结果的目的地存储器位置。第一组电子电路确定指示要执行的表格数据操作操作的控制信息，并将该控制信息传输到第二组电子电路。基于控制信息，第二组电子电路从源存储器位置检索表格数据并应用表格数据操作操作以生成数据操纵结果，并且使数据操作结果存储在目的地存储器位置。

2. 发明申请

WO2009058448A1 DOCUMENT FIDELITY WITH BINARY XML STORAGE 审中-公开
标题翻译：具有二进制XML存储的文档清晰度
公开(公告)号：WO2009058448A1
公开(公告)日：2009-05-07
申请号：PCT/US2008/071492
申请日：2008-07-29
申请人： ORACLE INTERNATIONAL CORPORATION , STHANIKAM, Balasubramanyam , CHANDRASEKAR, Sivasankaran , AGARWAL, Nipun , IDICULA, Sam , MEDI, Vijay
发明人： STHANIKAM, Balasubramanyam , CHANDRASEKAR, Sivasankaran , AGARWAL, Nipun , IDICULA, Sam , MEDI, Vijay
IPC分类号： G06F17/30
CPC分类号： G06F17/30569 , G06F17/30908
摘要： Techniques are provided for ensuring lexical fidelity when an XML document is stored in a binary format. Operations, on the XML data, that would cause the loss of lexical fidelity between the original XML document and the binary-encoded version of the XML document are not performed. Such operations include the removal of unnecessary whitespace characters, certain data type conversions, CRLF normalization, the "collapsing" of two-tag empty elements into a single tag empty element, and the replacing of entity references or numeric character references with another value. An XML schema, to which the XML document conforms, may indicate that the XML document is to be stored in a lexical fidelity mode. Additionally, or alternatively, the database statement that (when executed) causes the XML document to be stored in a binary format may so indicate.
摘要翻译：当XML文档以二进制格式存储时，提供了用于确保词汇保真度的技术。不会执行对原始XML文档和XML文档的二进制编码版本之间的词法保真度的XML数据的操作。这样的操作包括删除不必要的空白字符，某些数据类型转换，CRLF归一化，将双标签空元素“折叠”到单个标签空元素中，以及用另一个值替换实体引用或数字字符引用。 XML文档符合的XML模式可以指示XML文档将以词法保真模式存储。另外或替代地，数据库语句（当被执行时）导致XML文档以二进制格式存储可以这样表示。

3. 发明申请

WO2020242767A1 USING METAMODELING FOR FAST AND ACCURATE HYPERPARAMETER OPTIMIZATION OF MACHINE LEARNING AND DEEP LEARNING MODELS 审中-公开
公开(公告)号：WO2020242767A1
公开(公告)日：2020-12-03
申请号：PCT/US2020/032774
申请日：2020-05-14
申请人： ORACLE INTERNATIONAL CORPORATION
发明人： MOHARRER, Ali , VARADARAJAN, Venkatanathan , IDICULA, Sam , AGRAWAL, Sandeep , AGARWAL, Nipun
IPC分类号： G06N20/20 , G06N5/00 , G06N7/00 , G06N3/08
摘要： Herein are techniques that train regressor(s) to predict how effective would a machine learning model (MLM) be if rained with new hyperparameters and/or dataset. In an embodiment, for each training dataset, a computer derives, from the dataset, values for dataset metafeatures. The computer performs, for each hyperparameters configuration (HC) of a MLM, including landmark HCs: configuring the MLM based on the HC, training the MLM based on the dataset, and obtaining an empirical quality score that indicates how effective was said training the MLM when configured with the HC. A performance tuple is generated that contains: the HC, the values for the dataset metafeatures, the empirical quality score and, for each landmark configuration, the empirical quality score of the landmark configuration and/or the landmark configuration itself. Based on the performance tuples, a regressor is trained to predict an estimated quality score based on a given dataset and a given HC.

4. 发明申请

WO2019067957A1 ALGORITHM-SPECIFIC NEURAL NETWORK ARCHITECTURES FOR AUTOMATIC MACHINE LEARNING MODEL SELECTION 审中-公开
公开(公告)号：WO2019067957A1
公开(公告)日：2019-04-04
申请号：PCT/US2018/053515
申请日：2018-09-28
申请人： ORACLE INTERNATIONAL CORPORATION
发明人： AGRAWAL, Sandeep , IDICULA, Sam , VARADARAJAN, Venkatanathan , AGARWAL, Nipun
IPC分类号： G06N20/20
摘要： Techniques are provided for selection of machine learning algorithms based on performance predictions by trained algorithm-specific regressors. In an embodiment, a computer derives meta-feature values from an inference dataset by, for each meta-feature, deriving a respective meta-feature value from the inference dataset. For each trainable algorithm and each regression meta-model that is respectively associated with the algorithm, a respective score is calculated by invoking the meta-model based on at least one of: a respective subset of meta-feature values, and/or hyperparameter values of a respective subset of hyperparameters of the algorithm. The algorithm(s) are selected based on the respective scores. Based on the inference dataset, the selected algorithm(s) may be invoked to obtain a result. In an embodiment, the trained regressors are distinctly configured artificial neural networks. In an embodiment, the trained regressors are contained within algorithm-specific ensembles. Techniques are also provided for optimal training of regressors and/or ensembles.

5. 发明申请

WO2020068471A1 DISK DRIVE FAILURE PREDICTION WITH NEURAL NETWORKS 审中-公开
公开(公告)号：WO2020068471A1
公开(公告)日：2020-04-02
申请号：PCT/US2019/051316
申请日：2019-09-16
申请人： ORACLE INTERNATIONAL CORPORATION
发明人： KOCBERBER, Onur , SCHMIDT, Felix , RAGHAVAN, Arun , AGARWAL, Nipun , IDICULA, Sam , ZHOU, Guang-Tong , KUNAL, Nitin
IPC分类号： G06F11/00 , G06N3/02
摘要： Disk drive failure is predicted using a machine learning model. The framework involves receiving disk drive sensor attributes as training data, preprocessing the training data to select a set of enhanced feature sequences, and using the enhanced feature sequences to train a machine learning model to predict disk drive failures from disk drive sensor monitoring data. Prior to the training phase, the RNN LSTM model is tuned using a set of predefined hyper-parameters. The preprocessing, which is performed during the training and evaluation phase as well as later during the prediction phase, involves using predefined values for a set of parameters to generate the set of enhanced sequences from raw sensor reading. The enhanced feature sequences are generated to maintain a desired healthy/failed disk ratio, and only use samples leading up to a last-valid-time sample to honor a pre-specified heads-up-period alert requirement.

6. 发明申请

WO2014043366A2 OPTIMAL DATA REPRESENTATION AND AUXILIARY STRUCTURES FOR IN-MEMORY DATABASE QUERY PROCESSING 审中-公开
标题翻译：用于内存数据库查询处理的最佳数据表示和辅助结构
公开(公告)号：WO2014043366A2
公开(公告)日：2014-03-20
申请号：PCT/US2013/059486
申请日：2013-09-12
申请人： ORACLE INTERNATIONAL CORPORATION
发明人： IDICULA, Sam , MOORE, Kevin , GOLD, Brian , AGARWAL, Nipun , SEDLAR, Eric
IPC分类号： G06F17/30
CPC分类号： G06F17/30 , G06F17/30289 , G06F17/30312
摘要： A method for providing optimized data representation of relations for in-memory database query processing is disclosed. The method seeks to optimize the use of the available memory by encoding relations on which the in-memory database query processing is performed and by employing auxiliary structures to maintain performance. Relations are encoded based on data patterns in one or more attribute-columns of the relation and the encoding that is selected is suited to a particular type of data in the column. Members of a set of auxiliary structures are selected based on the benefit the structure can provide and the cost of the structure in terms of the amount of memory used. Encoding of the relations is performed in real-time while query processing occurs, using locks to eliminate conflicts between the query processing and encoding.
摘要翻译：公开了一种用于提供用于内存数据库查询处理的关系的优化数据表示的方法。该方法寻求通过对执行内存数据库查询处理的关系进行编码并通过采用辅助结构来维持性能来优化对可用存储器的使用。关系根据关系的一个或多个属性列中的数据模式进行编码，并且所选编码适合列中特定类型的数据。一组辅助结构的成员是根据结构可以提供的好处和结构在所用内存数量方面的成本来选择的。关系的编码在查询处理发生时实时执行，使用锁来消除查询处理和编码之间的冲突。

7. 发明申请

WO2004044738A2 TECHNIQUES FOR SUPPORTING APPLICATION-SPECIFIC ACCESS CONTROLS WITH A SEPARATE SERVER 审中-公开
标题翻译：用独立服务器支持应用特定访问控制的技术
公开(公告)号：WO2004044738A2
公开(公告)日：2004-05-27
申请号：PCT/US2003/035494
申请日：2003-11-06
申请人： ORACLE INTERNATIONAL CORPORATION
发明人： IDICULA, Sam , AGARWAL, Nipun , MURTHY, Ravi
IPC分类号： G06F9/40
CPC分类号： G06F21/629 , G06F21/6218 , G06F21/6227 , G06F2221/2113 , G06F2221/2141 , Y10S707/956
摘要： Techniques for supporting access controls on application-specific operations performed by an application include receiving first data at a server distinct from the application. The first data describes a first set of privileges for performing a first set of application-specific operations. Second data is also received at the server. The second data associates a first user of the application with a privilege in the first set of privileges. In response to receiving a request at the server from the application, it is determined whether a particular user may have the application perform a particular application-specific operation based on the first data and the second data. The request indicates the particular user and the particular application-specific operation. A response is sent to the application. The response indicates whether the particular user may have the application perform the particular application-specific operation.
摘要翻译：用于支持由应用执行的特定于应用的操作的访问控制的技术包括在不同于该应用的服务器处接收第一数据。第一个数据描述了执行第一组应用程序特定操作的第一组权限。服务器也收到第二个数据。第二数据将应用程序的第一用户与第一组权限中的特权相关联。响应于从应用程序在服务器处接收到请求，确定特定用户是否可以使应用程序基于第一数据和第二数据执行特定的特定应用程序操作。该请求指示特定用户和特定应用程序特定的操作。响应发送到应用程序。响应指示特定用户是否可以使应用程序执行特定的应用程序特定操作。

8. 发明申请

WO2020214396A1 AUTOMATIC FEATURE SUBSET SELECTION BASED ON META-LEARNING 审中-公开
公开(公告)号：WO2020214396A1
公开(公告)日：2020-10-22
申请号：PCT/US2020/025646
申请日：2020-03-30
申请人： ORACLE INTERNATIONAL CORPORATION
发明人： KARNAGEL, Tomas , IDICULA, Sam , MOGHADAM, Hesam Fathi , AGARWAL, Nipun
IPC分类号： G06N20/20
摘要： The present invention relates to dimensionality reduction for machine learning (ML) models. Herein are techniques that individually rank features and combine features based on their rank to achieve an optimal combination of features that may accelerate training and/or inferencing, prevent overfitting, and/or provide insights into somewhat mysterious datasets. In an embodiment, a computer ranks features of datasets of a training corpus. For each dataset and for each landmark percentage, a target ML model is configured to receive only a highest ranking landmark percentage of features, and a landmark accuracy achieved by training the ML model with the dataset is measured. Based on the landmark accuracies and meta-features values of the dataset, a respective training tuple is generated for each dataset. Based on all of the training tuples, a regressor is trained to predict an optimal amount of features for training the target ML model.

9. 发明申请

WO2020081747A1 MINI-MACHINE LEARNING 审中-公开
公开(公告)号：WO2020081747A1
公开(公告)日：2020-04-23
申请号：PCT/US2019/056625
申请日：2019-10-17
申请人： ORACLE INTERNATIONAL CORPORATION
发明人： AGRAWAL, Sandeep , VARADARAJAN, Venkatanathan , IDICULA, Sam , AGARWAL, Nipun
IPC分类号： G06N5/00 , G06N20/00 , G06N3/08 , G06N7/00
摘要： Techniques are described for generating and applying mini-machine learning variants of machine learning algorithms to save computational resources in tuning and selection of machine learning algorithms. In an embodiment, at least one of the hyper-parameter values for a reference variant is modified to a new hyper-parameter value thereby generating a new variant of machine learning algorithm from the reference variant of machine learning algorithm. Based on the cost metric of the new variant of machine learning algorithm and comparing the performance score for the new and reference variants, the system determines whether the modified reference machine algorithm is the mini-machine learning algorithm that is computationally less costly than the reference variant of machine learning algorithm but closely tracks the accuracy thereof

10. 发明申请

WO2020081229A1 AUTOMATIC FEATURE SUBSET SELECTION USING FEATURE RANKING AND SCALABLE AUTOMATIC SEARCH 审中-公开
公开(公告)号：WO2020081229A1
公开(公告)日：2020-04-23
申请号：PCT/US2019/054118
申请日：2019-10-01
申请人： ORACLE INTERNATIONAL CORPORATION
发明人： KARNAGEL, Tomas , IDICULA, Sam , AGARWAL, Nipun
IPC分类号： G06N20/20 , G06N5/00 , G06N3/04 , G06N3/08 , G06N7/00 , G06N20/10 , G06N3/12
摘要： The present invention relates to dimensionality reduction for machine learning (ML) models. Herein are techniques that individually rank features and combine features based on their rank to achieve an optimal combination of features that may accelerate training and/or inferencing, prevent overfitting, and/or provide insights into somewhat mysterious datasets. In an embodiment, a computer calculates, for each feature of a training dataset, a relevance score based on: a relevance scoring function, and statistics of values, of the feature, that occur in the training dataset. A rank based on relevance scores of the features is calculated for each feature. A sequence of distinct subsets of the features, based on the ranks of the features, is generated. For each distinct subset of the sequence of distinct feature subsets, a fitness score is generated based on training a machine learning (ML) model that is configured for the distinct subset.

你已经成功收藏专利！

检索式保存成功!

IPRDB

热门服务

关于我们

友情链接

联系方式