Abstract: In one embodiment, a computing system may receive, by a just-in-time compiler, a plurality of bytecode to dynamically modify prior to executing. The computing system may extract, using the just-in-time compiler, sequences of one or more operations from the plurality of bytecode. The computing system may generate, using the just-in-time compiler an FX graph based on the sequences of the one or more operations. The computing system may compile, using a user-defined compiler, the FX graph into a compiled function. The computing system may execute the plurality of bytecode based at least on the compiled function.

Abstract: A cloud-based framework dynamically utilizes a distributed pool of accelerators to parallelize calculations of physical simulation (physics) solver code partitioned across multiple accelerators and compute nodes of one or more virtual data centers in a virtualized computing environment. Multi-level partitioning logic of the framework partitions an input data set of the physics solver code into code groups configured to run on the accelerators using a “hardware agnostic” software layer that abstracts differences in processing architectures to allow targeting of different types of accelerators. A predictive scheduler interacts with the multi-level partitioning logic to locate and predictively reserve the accelerators within the pool, dynamically access and utilize the accelerators when needed, and then promptly release them upon completion of the calculations.

Abstract: Systems and techniques are provided for a tree structure for byzantine fault tolerance. A validator computing device may receive a candidate block from a block producer. The block producer may be a second validator computing device of the distributed network. The validator computing device may determine whether or not the candidate block is valid. The validator computing device may sign an indication of whether or not the candidate block is valid to generate a signature. The validator computing device may send the signature to a third validator computing device that is a sibling node to the validator computing device. The validator computing device may receive a second signature from the third validator computing device. The validator computing device may combine the signature and the second signature to generate a combine signature if the signature and the second signature agree.

Abstract: A training set may be created to train a machine learning engine, such as an artificial neural network (ANN), to value a target domain name using data from previously sold domain names. The training set may comprise a plurality of word features vector of real numbers (information related to the words or tokens within the sold domain names), a plurality of word embedding vector of real numbers (word embedding of the words within the sold domain names), a plurality of context embedding vector of real numbers (sale context, i.e., location and date of a sale of a sold domain name), a plurality of DNS embedding vector of real number (DNS information of the sold domain name) and/or a plurality of domain name features vector of real numbers (data regarding the sold domain name). The ANN may then be trained on the training set, using the methods of gradient descent and back propagation, to value a target domain name.

Abstract: Systems and techniques are provided for a tree structure for byzantine fault tolerance. A validator computing device may receive a candidate block from a block producer. The block producer may be a second validator computing device of the distributed network. The validator computing device may determine whether or not the candidate block is valid. The validator computing device may sign an indication of whether or not the candidate block is valid to generate a signature. The validator computing device may send the signature to a third validator computing device that is a sibling node to the validator computing device. The validator computing device may receive a second signature from the third validator computing device. The validator computing device may combine the signature and the second signature to generate a combine signature if the signature and the second signature agree.

Abstract: Systems and techniques are provided for a tree structure for byzantine fault tolerance. A validator computing device may receive a candidate block from a block producer. The block producer may be a second validator computing device of the distributed network. The validator computing device may determine whether or not the candidate block is valid. The validator computing device may sign an indication of whether or not the candidate block is valid to generate a signature. The validator computing device may send the signature to a third validator computing device that is a sibling node to the validator computing device. The validator computing device may receive a second signature from the third validator computing device. The validator computing device may combine the signature and the second signature to generate a combine signature if the signature and the second signature agree.

Abstract: Illustrative embodiments improve upon prior machine learning techniques by introducing an additional classification layer that mimics human visual pattern recognition. Building upon classification passes that extract contextual information, illustrative embodiments look for hints of high-level semantic categorization that manifest as visual artifacts in the document, such as font family, font weight, text color, text justification, white space, or CSS class name. An improved lightweight markup language enables display of machine-categorized tokens on a screen for human correction, thereby providing ground truths for further machine classification.

Abstract: Systems and methods of the present invention provide for one or more server computers communicatively coupled to a network and configured to: access a source code for each of a plurality of web pages within a website hosted on the server computer; identify, within the source code of each of the plurality of web pages, a plurality of hyperlinks for navigating to at least one of the plurality of web pages; generate a plurality of link groups each comprising at least one common hyperlink between the plurality of hyperlinks; aggregate the plurality of link groups into a unique link group wherein each of the plurality of hyperlinks appears in only one link group; and determine an order of hyperlinks within the unique link group based on an original order of the plurality of hyperlinks.

Abstract: A system and method for generating web page layouts is presented. A computer server is configured to retrieve blocks of text for a web page. Each block of text is rendered into columns of text. A presentation score is calculated for each column of text. Populated web page layouts are generated for each web page layout in a plurality of web page layouts by determining a number of columns in the web page layout, for each block of text, positioning a rectangle having the same width and height as one of the plurality of columns of text for each block of text into a column in the web page layout to generate a populated web page layout, and calculating, without user input, a rendering score for the populated web page layout. One of the populated web page layouts is used to render a web page.

Abstract: Systems and methods of the present invention provide for server(s) assigning section or list item classifications to price list or business data extracted from a website. The server routes each new task verifying the classification to a crowd worker, and the server receives a completed. The server calculates a crowd worker score for each crowd worker based on each worker's quality scores according to the worker's review of the classifications on a worker user interface. The server generates a quality model for predicting a task quality score for the task, according to an error score for the crowd worker. If the error score in the quality model is below a predetermined threshold, the server transmits the completed task to a task reviewer's client for review.

Abstract: Systems and methods of the present invention provide for one or more server computers configured to assign section or list item classifications to price list or business data extracted from a website. The server calculates a crowd worker score for each of a plurality of crowd workers based on each worker's quality and speed scores for tasks reviewing the classifications on a worker user interface. If a crowd worker score for a worker is below a crowd worker quality threshold, each new task is routed to the worker, and the received task, when completed, is routed to a worker whose crowd worker score is above the crowd worker quality threshold for review.

Abstract: Systems and methods of the present invention provide for one or more server computers configured to assign section or list item classifications to price list or business data extracted from a website. The server assigns section or list item classifications to price list or business data extracted from a website. The server calculates a crowd worker score for each of a plurality of crowd workers based on each worker's quality and speed scores for tasks reviewing the classifications on a worker user interface. If a crowd worker score for a worker is below a crowd worker quality threshold, each new task is routed to the worker, and the received task, when completed, is routed to a worker whose crowd worker score is above the crowd worker quality threshold for review.

Abstract: A system and method for document classification are presented. An input document is received (e.g., by at least one server communicatively coupled to a network). A plurality of features are identified in the input document. The plurality of features include sequences of text extracted from the input document. A feature vector of the input document is generated based upon the sequences of text, and the feature vector of the input document is compared to each of a plurality of signature vectors to determine a primary type of the input document. The primary type of the input document is stored into a storage system in communication with the at least one server.

Abstract: A system and method for data classification are presented. A plurality of training tokens are identified by at least one server communicatively coupled to a network. Each training token includes a token retrieved from a content source and a classification of the token. For each training token in the plurality of training tokens, a plurality of n-gram sequences are identified, a plurality of features for the plurality of n-gram sequences are generated, and first training data is generated using the token retrieved from the content source, the plurality of features, and the classification of the token. A first classifier is trained with the first training data, and the first classifier is stored into a storage system in communication with the at least one server.

Abstract: Illustrative embodiments improve upon prior machine learning techniques by introducing an additional classification layer that mimics human visual pattern recognition. Building upon classification passes that extract contextual information, illustrative embodiments look for hints of high-level semantic categorization that manifest as visual artifacts in the document, such as font family, font weight, text color, text justification, white space, or CSS class name. An improved lightweight markup language enables display of machine-categorized tokens on a screen for human correction, thereby providing ground truths for further machine classification.

Abstract: Illustrative embodiments improve upon prior machine learning techniques by introducing an additional classification layer that mimics human visual pattern recognition. Building upon classification passes that extract contextual information, illustrative embodiments look for hints of high-level semantic categorization that manifest as visual artifacts in the document, such as font family, font weight, text color, text justification, white space, or CSS class name. An improved lightweight markup language enables display of machine-categorized tokens on a screen for human correction, thereby providing ground truths for further machine classification.

Abstract: A network caching system has a multi-protocol caching filer coupled to an origin server to provide storage virtualization of data served by the filer in response to data access requests issued by multi-protocol clients over a computer network. The multi-protocol caching filer includes a file system configured to manage a sparse volume that “virtualizes” a storage space of the data to thereby provide a cache function that enables access to data by the multi-protocol clients. To that end, the caching filer further includes a multi-protocol engine configured to translate the multi-protocol client data access requests into generic file system primitive operations executable by both the caching filer and the origin server.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for runtime language-independent sandboxing of software. In one aspect, a system implements an extended Software Fault Isolation (SFI) software sandboxing system configured to provide a user-mode program interface for receiving runtime requests for modifying verifiably safe executable machine code. Requests can include dynamic code creation, dynamic code deletion, and atomic modification of machine code instructions. A runtime modification of a verifiably safe executable memory region is made in response to each received runtime request, and code within the modified memory region has a guarantee of safe execution.

Abstract: A system and method for automatically determining an amount of review a crowd-sourcing task needs after an initial review has been completed by a processing worker. An evaluation metric is automatically assigned to the work performed by the processing worker to determine the appropriate amount of human review required for a particular task. The evaluation metric may be calculated by accessing and evaluating a plurality of transaction categories related, but not limited to, worker characteristics, document characteristics and processing characteristics. Additionally, the evaluation metric may be used to determine compensation of the processing worker and whether a promotion or demotion is necessary. The system is also capable of balancing individual workloads based upon the evaluation metric.

Abstract: A method for generating a website includes obtaining a seed input associated with an entity. The seed input may include one or more keywords, such as a business name. Obtaining the seed input may include receiving the seed input from the user, or the seed input may be obtained without input from the user. The seed input is used to identify the entity. The method further includes retrieving, using at least one of the seed input and the identification of the entity, content relevant to the entity from one or more data stores. Retrieving the content may include using one or more categories relevant to the entity to identify the content. The website is generated without an input from the entity, and includes at least a portion of the content. Generating the website may include identifying a template having a plurality of content regions for containing the content.