Abstract: System and method for creating clips of larger audio streams based on very accurate in point and out point and for searching for the clips. The system and method use byte signatures in MP3 or media files to accurately and efficiently find a specific cue point, such as for example, the beginning of a sentence in a podcast. The system and method are configured to search and locate a given clip even if that sequence moves inside the stream, such as for example, if the audio file adds additional content like advertisements or other audio sequence, the hash representation of the cue point can still be found with millisecond accuracy.

Abstract: A data storage device whose controller is configured to apply a hash function to a logical address specified in a received host request to obtain a first portion of the corresponding physical address (e.g., the channel number or channel and die numbers). This feature of the controller enables the L2P table stored in the DRAM associated with the controller to have physical-address entries that contain therein only complementary second portions of the physical addresses, but not the first portions. Such shorter physical-address entries in the L2P table enable a corresponding beneficial reduction in the size of the DRAM and can further be leveraged to have optimized and aligned access to the L2P table in the DRAM.

Abstract: Disclosed are a system comprising a computer-readable storage medium storing at least one program, and a computer-implemented method for event messaging over a network. An identification module receives first data identifying consumer devices available to receive event messages linked to a topic. The identification module receives second data identifying a producer device available to provide event messages linked to the topic. A provisioning module links a plurality of values to respective consumer devices that are linked to the topic. A scheduler module accesses a first data message linked to the topic. The first data message includes a key value. The scheduler module provides the first data message to a selected one of the consumer devices based on a comparison of the key value and the plurality of values of the respective consumer devices.

Abstract: The present systems, devices, and methods relate to managing shared vehicle access. Vehicle access can be provided to a candidate user based on whether the user is included in a list of at least one user associated with a respective virtual key. Virtual keys can be created, or lists of at least one user can be updated to include the candidate user, to in turn provide vehicle access. Virtual keys can be stored in a plurality of virtual key slots, with vehicle access being provided based on a virtual key stored in an active key slot. Status reports can be sent to a reservation management device, for updating virtual keys or lists of at least one user.

Abstract: In a method for identifying visually similar media content items, perceptual hashes for video frames of media content items are received. The perceptual hashes are compared for at least a portion of video frames. Based on the comparing the perceptual hashes for at least a portion of video frames, it is determined whether media content items are matching. The media content items indicated as matching are grouped.

Abstract: An example method of performing searches by comparing regions includes identifying a search point contained in geo-located data. The search point is nearest in distance to a query location. A search region associated with the search point is determined. The search region includes neighbor points contained in the geo-located data that are nearest the search point. Centroids corresponding respectively to the plurality of neighbor points of the search region are determined. The centroids of the search region are compared to other centroids corresponding to other regions of the geo-located data. The number of other centroids that match with the plurality of centroids are counted, and the counts are placed in descending order, to list the other regions having the highest similarity first. An ordered list of the other regions is generated based on their respective similarities to the query region.

Abstract: A key-value storage device includes a non-volatile memory and a controller. The key-value storage device is configured to communicate with a host and includes a controller and a non-volatile memory. The controller is configured to receive, from the host, a first command including a key, a first chunk among multiple chunks included in a value corresponding to the key, and a first chunk index corresponding to the first chunk. The controller is configured to store mapping information of the first chunk in a mapping table based on the key and the first chunk index. The non-volatile memory is configured to store the key and the first chunk based on the mapping information.

Abstract: A method includes determining that an amount of available space in a first memory used by a set of relational queries is insufficient for a query, wherein the query comprises a join operation. A first partition of a set of partitions is identified, wherein the first partition possesses a smallest available probe memory size of the set of partitions and a build memory size greater than or equal to a threshold memory size, wherein the threshold memory size is a percentage of a maximum build memory size, and the largest partition of the set of partitions has the maximum build memory size. The first partition is copied from the first memory to a second memory.

Abstract: The present disclosure involves systems, software, and computer implemented methods for improved design and implementation of data access metrics for automated physical database design. An example method includes identifying a database workload for which index advisor access counters are to be tracked. Each SQL statement in the database workload is executed. For each SQL statement, attribute sets are determined for which a selection predicate filters a result for an SQL statement. An output cardinality of each selection predicate is determined. A logarithmic counter for an attribute set corresponding to the selection predicate is determined based on the output cardinality of the selection predicate. The determined logarithmic counter is incremented. Respective values for logarithmic counters of the determined attributes are provided to an index advisor. The index advisor determines attribute sets for which to propose an index based on the logarithmic counters of the respective attribute sets.

Abstract: A system for training a neural-network-based floating-point-to-binary feature vector encoder preserves the locality relationships between samples in an input space over to an output space. The system includes a neural network under training and a probability distribution loss function generator. The neural network has floating-point inputs and floating-point pseudo-bipolar outputs. The generator compares an input probability distribution constructed from floating-point cosine similarities of an input space and an output probability distribution constructed from floating-point pseudo-bipolar pseudo-Hamming similarities of an output space. The system includes a proxy vector set generator to take a random sampling of vectors from training data for a proxy set, a sample vector selector to select sample vectors from the training data and a KNN vector set generator to find a set of k nearest neighbors closest to each sample vector from said proxy set for a reference set.

Abstract: A computer implemented method for classifying at least one source dataset of a computer system. The method may include providing a plurality of associated reference tables organized and associated in accordance with a reference storage model in the computer system. The method may also include calculating, by a data classifier application of the computer system, a first similarity score between the source dataset and a first reference table of the reference tables based on common attributes in the source dataset and a join of the first reference table with at least one further reference table of the reference tables having a relationship with the first reference table. The method may further include classifying, by the data classifier application, the source dataset by determining using at least the calculated first similarity score whether the source dataset is organized as the first reference table in accordance to the reference storage model.

Abstract: The sport data tracking systems available today are based on specialized hardware to detect and track targets on the field. While effective, implementing and maintaining these systems pose a number of challenges, including high cost and need for close human monitoring. On the other hand, the sports analytics community has been exploring human computation and crowdsourcing in order to produce tracking data that is trustworthy, cheaper and more accessible. However, state-of-the-art methods require a large number of users to perform the annotation, or put too much burden into a single user. Example methods, systems and user interfaces that facilitate the creation of tracking data sequences of events (e.g., plays of baseball games) by warm-starting a manual annotation process using a vast collection of historical data are described.

Abstract: The technology disclosed generates a reference array of variant data for locations that are shared between read results which are to be compared, and generates hashes over a selected pattern length of positions in the reference array to independently produce non-unique window hashes for base patterns in the read results. It then selects for comparison window hashes that occur less than a ceiling number of times and compares the selected window hashes to identify common window hashes between the read results. It then determines a similarity measure for the read results based on the common window hashes.

Abstract: Disclosed is an approach to implement an architecture and methodology that allows rapid access to provisioned software without requiring the entirety of the software distribution to be transferred to the target system in an upfront manner. Instead, a multi-tier architecture is used that allows the software to be provisioned and accessed with efficient access and reads of distribution materials from one or more remote storage locations.

Abstract: The technology disclosed generates a reference array of variant data for locations that are shared between read results which are to be compared, and generates hashes over a selected pattern length of positions in the reference array to independently produce non-unique window hashes for base patterns in the read results. It then selects for comparison window hashes that occur less than a ceiling number of times and compares the selected window hashes to identify common window hashes between the read results. It then determines a similarity measure for the read results based on the common window hashes.

Abstract: A string, identifying an item to be assigned to a physical resource, is hashed to obtain a numeric hash value. The numeric hash value is downscaled to obtain a bucket identifier that identifies a bucket that will hold the numeric hash value. The bucket is then deterministically mapped to a physical resource so that it can be retrieved without accessing a stored data structure representative of the mapping.

Abstract: The present systems, devices, and methods relate to managing shared vehicle access. Vehicle access can be provided to a candidate user based on whether the user is included in a list of at least one user associated with a respective virtual key. Virtual keys can be created, or lists of at least one user can be updated to include the candidate user, to in turn provide vehicle access. Virtual keys can be stored in a plurality of virtual key slots, with vehicle access being provided based on a virtual key stored in an active key slot. Status reports can be sent to a reservation management device, for updating virtual keys or lists of at least one user.

Abstract: Disclosed is a system, method, and computer program product to efficiently process multi-set operations in a database system. An approach is described to perform a group-by operation with a counter to efficiently process such queries. Techniques are described to optimize multi-set operations into regular-set operations.

Abstract: A method and/or system of managing metadata are disclosed that include connecting a source data storage system (DSS) that stores both data and metadata to a metadata management platform (MMP); scanning metadata records onto the MMP from the DSS; storing metadata attributes for at least one of the group consisting of directories and buckets on the DSS in a look-up table on the MMP; and adding updated metadata attributes to the look-up table on the MMP for each subsequent scan of the DSS.

Abstract: Examples described herein relate to a neural network whose weights from a matrix are selected from a set of weights stored in a memory on-chip with a processing engine for generating multiply and carry operations. The number of weights in the set of weights stored in the memory can be less than a number of weights in the matrix thereby reducing an amount of memory used to store weights in a matrix. The weights in the memory can be generated in training using gradients from back propagation. Weights in the memory can be selected using a tabulation hash calculation on entries in a table.

Abstract: An information processing apparatus includes: a memory; and a processor coupled to the memory and configured to: receive personal data related to a personal data originator; receive agreement information on a processing method for the personal data agreed between the personal data originator and a personal data user who uses the personal data; process the personal data by the processing method defined in the agreement information; and attach a digital signature to processed data and output the processed data.

Abstract: Object scriptability methods and system are described herein. The method includes generating a graph-based data structure including a plurality of nodes, where each node from the plurality of nodes represents a respective data-analysis object in a data analysis system, where each node from the plurality of nodes is connected to at least one other node from the plurality of nodes by an edge, where the edge represents a relationship between the respective objects in the data analysis system, and where generating the graph-based data structure includes receiving a high-level language script describing at least one data-analysis object, and generating at least one node from the plurality of nodes in accordance with the high-level language script.

Abstract: A schema for a dataset is identified by identifying a dataset comprising data and relationships between data pairs. An original schema is identified for the dataset. This original schema comprises an organizational structure. An initial fit between the dataset and the original schema is determined. The initial fit quantifying a conformity of the data in the dataset to the organizational structure of the original schema. A plurality of additional schemas are identified. Each additional schema is a distinct organizational schema. The dataset is partitioned into a plurality of subsets. Each subset comprises a modified fit quantifying a modified conformity of subset data in each subset to one of the original schema and the additional schemas. The modified fit is greater than the original fit.

Abstract: Aspects of the present invention are directed to system and methods for optimizing identification of locations within a search area using hash values. A hash value represents location information in a single dimension format. Computing points around some location includes calculating an identification boundary that surrounds the location of interest based on the location's hash value. The identification boundary is expanded until it exceeds a search area defined by the location and a distance. Points around the location can be identified based on having associated hash values that fall within the identification boundary. Hashing operations let a system reduce the geometric work (i.e. searching inside boundaries) and processing required, by computing straightforward operations on hash quantities (e.g. searching a linear range of geohashes), instead of, for example, point to point comparisons.

Abstract: Techniques and systems for reducing data stored on a block processing storage system are described. A losslessly reduced representation of a data block can include references to one or more prime data element blocks, and optionally a description of a reconstitution program which, when applied to the one or more prime data element blocks, results in the data block.

Abstract: During operation, a key-value storage system can receive a request to write data to a data region in a first non-volatile memory. The system can determine a key associated with the data and the key can correspond to an entry in a data structure maintained by a volatile memory the storage system. In response to determining the key, the system can write the data to the data region in the first non-volatile memory. The system can update in the volatile memory the data structure entry corresponding to the key with a physical location in the data region of the first non-volatile memory to which the data is written. The system can write the data structure update to a second non-volatile memory. The system can then store a snapshot of the data structure in a metadata region associated with the first non-volatile memory, thereby allowing persistent storage of the data structure.

Abstract: Systems and methods for spilling data for hash joins are described. An example method includes determining an amount of available space in a first memory used by a set of relational queries is insufficient for a first relational join query. The first relational join query comprises a join operation. The method also includes determining a set of build memory sizes and a set of probe memory sizes for a set of partitions for the set of relational queries. The method further includes identifying a first partition of the set of partitions based on the set of probe memory sizes and the set of build memory sizes. The method further includes copying the first partition from the first memory to a second memory, wherein the first partition comprises a first build portion and a first probe portion.

Abstract: The technology disclosed generates a reference array of variant data for locations that are shared between read results which are to be compared, and generates hashes over a selected pattern length of positions in the reference array to independently produce non-unique window hashes for base patterns in the read results. It then selects for comparison window hashes that occur less than a ceiling number of times and compares the selected window hashes to identify common window hashes between the read results. It then determines a similarity measure for the read results based on the common window hashes.

Abstract: A sharded, permissioned, distributed ledger may reduce the amount of work and communication required by each participant, thus possibly avoiding scalability bottlenecks that may be inherent in previous distributed ledger implementations and possibly enabling the use of additional resources to translate to increased throughput. A sharded, permissioned, distributed ledger may be made up of multiple shards, each of which may also be a distributed ledger and which may operate in parallel. Participation within a sharded, permissioned, distributed ledger may be allowed only with permission of an authority. A sharded, permissioned, distributed ledger may include a plurality of nodes, each including a dispatcher configured to receive transaction requests from clients and to forward received requests to verifiers configured to append transactions to individual ones of the shards.

Abstract: The technology describes performing garbage collection while data writes are occurring, which can lead to a conflict in that a new reference to an otherwise non-referenced candidate object for garbage collection is written after the non-referenced candidate object is detected. In one example implementation, orphaned binary large objects (BLOBs) that are not referenced by a descriptor file and are beyond a certain age are detected and deleted via an object references table traversal as part of garbage collection. Before reclaiming a deleted BLOB's capacity, a background process operates to restore the deleted BLOB if a new descriptor file reference to the BLOB was written during the object references table traversal. Capacity is only reclaimed after the object references table traversal and the background processing completes, for those BLOBs that were deleted and had not been restored.

Abstract: A deduplication engine maintains a deduplication hash table containing a plurality of fingerprint entries identifying a track of current data and a hash value computed from the data currently stored at the identified track. A hash table manager implements a dynamic deduplication hash table management process, that includes a fingerprint entry purge process and a fingerprint entry re-entry process. The purge process is implemented on the deduplication hash table to remove selected fingerprint entries of tracks of data currently stored on the managed drives from the deduplication hash table based on a first set of Thin Device (TDev) selection rules and a second set of track selection purge rules. The re-entry process is implemented on the deduplication hash table to re-enter selected fingerprint entries into the deduplication hash table based on a third set of TDev selection rules and a fourth set of track selection re-entry rules.

Abstract: A method of securing media-based transactions and contracts on a distributed ledger network comprising receiving a first media file, obtaining a fingerprint of the first media file, generating a non-cryptographic hash value for the fingerprint of the first media file, generating a block for the distributed ledger network comprising a block header comprising previous hash, a state root hash, a receipts root hash, and a transaction list root hash, defining block header hashes, recording the non-cryptographic fingerprint hash value for the fingerprint of the first media file to the block and recording the block to the distributed ledger network.

Abstract: Example methods and systems for indexing fingerprints are described. Fingerprints may be made up of sub-fingerprints, each of which corresponds to a frame of the media, which is a smaller unit of time than the fingerprint. In some example embodiments, multiple passes are performed. For example, a first pass may be performed that compares the sub-fingerprints of the query fingerprint with every thirty-second sub-fingerprint of the reference material to identify likely matches. In this example, a second pass is performed that compares the sub-fingerprints of the query fingerprint with every fourth sub-fingerprint of the likely matches to provide a greater degree of confidence. A third pass may be performed that uses every sub-fingerprint of the most likely matches, to help distinguish between similar references or to identify with greater precision the timing of the match. Each of these passes is amenable to parallelization.

Abstract: Methods and systems are provided for identifying unique devices and/or unique users of a web-based system within constraints of an external application. In some embodiments the process comprises receiving request information from a client device at an application server, using such information to form a device fingerprint, and attempting to match the newly-formed fingerprint with exactly one fingerprint stored in a database of previously-formed fingerprints. Some embodiments utilize a two-stage Locality-Sensitive Hash query technique. The client device fingerprint may be converted into a series of LSH values which may be used to find a matching fingerprint. A first stage may query input LSH values against LSH values in a data store, and a second stage may query LSH values temporarily held in volatile memory, thereby minimizing network traffic and reducing a total process time.

Abstract: Techniques for message deduplication are described. In one embodiment, an apparatus may comprise a client front-end component operative to receive an incoming message at a messaging system from a messaging client on a client device; extract a deduplication token from the incoming message; store the incoming message in a message queue where examining the deduplication token determines that the incoming message does not duplicate any existing message of the messaging system; and discard the incoming message where examining the deduplication token determines that the incoming message does duplicate an existing message; and a message deduplication component operative to examine the deduplication token to determine whether the incoming message duplicates any existing message of the messaging system. Other embodiments are described and claimed.

Abstract: Disclosed is a system for restrictive clustering of datapoints. The system comprises server arrangement that acquires data record for performing clustering operation, determines datapoints for the data record, plots the datapoints in a multi-dimensional space, determines a cluster threshold, and performs a first iteration of clustering on the datapoints plotted in the multi-dimensional space, determines a segment threshold for the datapoints plotted in the multi-dimensional space, derives boundary conditions for determining segments based on the segment threshold and superimposes the boundary conditions corresponding to each of the segments based on the segment threshold onto the first iteration of clustering. Moreover, the server arrangement re-iterates the first iteration of clustering to obtain a second iteration of clustering, wherein the second iteration of clustering has an error value lower than an error value for the first iteration of clustering.

Abstract: The invention uses document retrieval to explain to a human user the properties of a query object that are revealed by a machine learning procedure, lending interpretability to the procedure. A query object is compared to reference objects by transforming the query object and reference objects into representative tokens. Reference objects with many tokens in common with the query object are returned as relevant result objects by a document retrieval system. The token representation furthermore admits comparison between features of the query object and matched features of the reference object or between the query object and groups of reference objects having common features, thus emphasising characteristics of the query and reference objects of semantic importance to the user based on the intention of their search. Embodiments include retrieval of 2-dimensional or 3-dimensional images, audio clips, and text.

Abstract: According to one embodiment, provided is a storage management system having a processor and managing a plurality of storages, in which the storage has a plurality of types of components, and performs redundantization of a data to be stored by combining the plurality of components, the storage management system has component management information indicating redundancy set for the component itself, and the processor sets a method of data protection of the storage based on the component management information and a request for the data protection for the data stored in the storage.

Abstract: A key-value storage device includes a non-volatile memory and a controller. The key-value storage device is configured to communicate with a host and includes a controller and a non-volatile memory. The controller is configured to receive, from the host, a first command including a key, a first chunk among multiple chunks included in a value corresponding to the key, and a first chunk index corresponding to the first chunk. The controller is configured to store mapping information of the first chunk in a mapping table based on the key and the first chunk index. The non-volatile memory is configured to store the key and the first chunk based on the mapping information.

Abstract: A method for cold matching by automatic content recognition includes receiving, at data processing hardware, broadcast fingerprints indicative of broadcast frames of a broadcast media stream. The method also includes receiving, at the data processing hardware, media device fingerprints indicative of media device frames of a media device stream at a corresponding media device. The method further includes determining, by the data processing hardware, that a search block comprising a plurality of media device fingerprints corresponding to sequential media device frames of the media device stream includes at least one media device fingerprint having a match correlation to at least one broadcast fingerprint. The match correlation is based on a Pearson correlation coefficient. The method also includes identifying, by the data processing hardware, the at least one media device fingerprint having the match correlation to the at least one broadcast fingerprint as corresponding to a match media device frame.

Abstract: A string, identifying an item to be assigned to a physical resource, is hashed to obtain a numeric hash value. The numeric hash value is downscaled to obtain a bucket identifier that identifies a bucket that will hold the numeric hash value. The bucket is then deterministically mapped to a physical resource so that it can be retrieved without accessing a stored data structure representative of the mapping.

Abstract: Examples disclosed herein are relevant to configurations of hash tables. An example hash table includes is configured to be placed into a contiguous block of shared memory and be used across multiple processes. The hash table can include a memory layout having two logical tables: a metadata table and a storage array. In addition, a storage bitmap can be used to determine available memory locations in the storage array. Disclosed configurations can be suited for use in high-performance computing applications.

Abstract: A storage unit (12) stores a hash table that corresponds to a predetermined field of a packet and has a set of a hash value from a value in the corresponding field and a pointer registered, and an ANY point table that corresponds to the predetermined field and has a pointer registered. A search unit (132) refers to a hash value calculated from a value in a field of a packet to be searched and to a pointer registered in the hash table and the ANY point table, to perform hash search for a rule corresponding to the packet to be searched.

Abstract: A computer implemented method for classifying at least one source dataset of a computer system. The method may include providing a plurality of associated reference tables organized and associated in accordance with a reference storage model in the computer system. The method may also include calculating, by a data classifier application of the computer system, a first similarity score between the source dataset and a first reference table of the reference tables based on common attributes in the source dataset and a join of the first reference table with at least one further reference table of the reference tables having a relationship with the first reference table. The method may further include classifying, by the data classifier application, the source dataset by determining using at least the calculated first similarity score whether the source dataset is organized as the first reference table in accordance to the reference storage model.

Abstract: A system for unloading tables of a database is provided. In some aspects, the system performs operations including determining that a number of accesses to a table occurring within a time period has satisfied an access threshold. The operations may further include identifying, in response to the determining, a first timestamp indicating a most recent access to the table. The operations may further include determining whether a difference between a current timestamp and the first timestamp satisfies a first time threshold. The operations may further include comparing, in response to the difference satisfying the first time threshold, a ratio of the difference and a size of the table to a ratio threshold. The operations may further include unloading, in response to satisfying the ratio threshold, the table. The operations may further include adjusting, based on the feedback, the first time threshold and/or the ratio threshold.

Abstract: Embodiments of the present disclosure relate to a method, apparatus, device, and medium for storing and querying data. The method for storing data includes: generating, for to-be-stored data, a primary index for indexing the data. The method further includes: writing the primary index arid the data into a data row in a data table. The method further includes: generating a secondary index for indexing the primary index based on a query rule of the data table and the data. In addition, the method further includes: writing the secondary index and the primary index into an index row in the data table.

Abstract: A data analysis device is configured to include an arithmetic operation device that determines whether or not each point data in the point data group of the existing clusters is a core of the cluster using a predetermined algorithm and calculates a predetermined re-verification determination data, recalculates an initial condition value of the clustering based on the re-verification determination data, specifies re-verification point data which is the point data that becomes no longer the core or becomes a new core by adding the point data, from the point data group to which the addition was made, based on the updated initial condition value obtained by the recalculation, and performs predetermined clustering processing in accordance with the re-verification point data.

Abstract: Systems and methods are described for implementing an “arc jump” technique in conjunction with bounded loads in consistent hashing. In general, bounded loads refers to limiting the ability of a single device within a distributed system to store data objects, such that when a request to store a new data object would otherwise be directed to that device, it is instead redirected to an alternative device. Redirecting all requests to a single alternative device can lead to cascading failures, as the alternative device must maintain its own load and that which has been redirected to it. Embodiments of the present disclosure address this by determining an alternative device on a per-object basis, such as by again hashing the object with an additional seed value. This distributes request from an overloaded device among all other devices of the distributed system, avoiding cascading failures.

Abstract: A user permission data query method which includes obtaining a first data table including staff identification numbers and departments corresponding to the staff identification numbers, and obtaining a second data table including a correspondence relationship among the staff identification numbers, roles, and administration authority information; obtaining, from the second data table, a plurality of data records having the same staff identification number and the same role, calculating an MD5 value corresponding to the staff identification number and the role; screening various MD5 values that are different from each other, and obtaining the management departments and the management staffs respectively corresponding to the various MD5 values obtaining a MD5 value corresponding to the permission query request and determining the management departments and the management staffs corresponding to the MD5 value as permission data of a user, when a permission query request is received.

Abstract: Changes made to a database table are accumulated, in durable storage, and snapshots of partitions of the table are obtained. For successive snapshots of a partition, the system accesses a previous snapshot, applies changes from the accumulated changes, and stores the updated snapshot to a durable data store. The accumulated changes and the successive partition snapshots are made available to restore the database to any point in time across a continuum between successive snapshots. Although each partition of the table may have a backup snapshot that was generated at a time different from when other partition snapshots were generated, changes from respective change logs may be selectively log-applied to distinct partitions of a table to generate an on-demand backup of the entire table at common point-in-time across partitions. Point-in-time restores of a table may rely upon a similar process to coalesce partition snapshots that are not aligned in time.