Abstract: An example method of message routing includes: receiving, by one or more processors, a request to send a message to a specified user of a plurality of users of a communication services platform; providing a user profile of the specified user to a communication channel selection model, wherein the user profile characterizes actions of one or more predefined action types that were performed by the specified user in response to receiving previous communications; identifying, based on the output of the communication channel selection model, a preferred communication channel for communicating with the specified user; determining, based on the preferred communication channel, a communication strategy for the specified user; and causing, pursuant to the communication strategy, a message to be sent to the specified user.

Abstract: An example method of message routing includes: receiving a request to send a message to a specified user of a plurality of users of a communication services platform; providing a user profile of the specified user to a send time optimization model, wherein the user profile characterizes actions of one or more predefined action types that were performed by the specified user in response to receiving previous communications; identifying, based on the output of the send time optimization model, a preferred time range for communicating with the specified user; and causing, within the preferred time range, a message to be sent to the specified user.

Abstract: The present invention proposed a super-resolution reconstruction device for Micro CT images of rat ankle fractures, comprising a rat ankle image preprocessing module, an HR LR image pair configuration module, a deep model module, and an image super-resolution reconstruction module. The aforementioned four modules were sequentially connected; the present invention also proposed an improved R2 RCAN model based on RCAN, which was a super-resolution reconstruction model grounded on self-attention mechanisms that enhanced the model's ability to extract features in multiple scales by incorporating Res2Net. Compared with other classic super-resolution models, the proposed R2 RCAN model achieved the best results.

Abstract: A non-uniform memory access (NUMA) architecture-based near-memory processing (NMP) emulator apparatus includes a host simulator module configured to perform a function of a host system using a non-uniform memory access (NUMA) node of a NUMA architecture, and one or more acceleration dual in-line memory module (AXDIMM) emulator modules, wherein each of the one or more AXDIMM emulator modules is configured to perform a function of a stream of an AXDIMM architecture using a NUMA node of the NUMA architecture.

Abstract: Devices and techniques are generally described for a speech processing routing architecture. First input data representing an input request may be received. First data including a semantic interpretation of the input request may be determined. Metadata of the first input data may be determined. The metadata may identify an entity associated with the input request. In some examples, a query may be sent to a first component. The query may include the metadata. In some examples, second data that identifies a first skill associated with the entity may be received from the first component. In various examples, the first skill may be selected for processing the first input data based at least in part on the first data and the second data.

Abstract: The present disclosure provides an intracranial artery stenosis detection method and system. The present disclosure obtains artery stenosis detection results based on a first maximum intensity projection (MIP) image and a second MIP image obtained by preprocessing a medical image by adopting a detection model based on an adaptive triplet attention module and generates an auxiliary report and visualization results according to target category information in the artery stenosis detection results. Therefore, the problem that existing manual interpretation methods are easily affected by the subjective experience of doctors and are time-consuming and laborious can be solved, thus improving accuracy and efficiency of intracranial artery stenosis detection.

Abstract: Methods, systems, and computer programs are presented for the determination of optimal communication scheduling. One method includes an operation for training a machine-learning program to generate a frequency model that determines a frequency for sending communications to users. The training utilizes training data defined by features related to user information and responses of users to previous communications to the users. The method further includes determining, by the frequency model and based on information about a first user, a first frequency for the first user. The first frequency identifies the number of communications to transmit to the first user per period of time. Further, the method includes operations for receiving a communication request to send one or more communications to the first user and determining send times for the one or more communications to the first user based on the first frequency. The communications are sent at the determined send times.

Abstract: Methods, systems, and computer programs are presented for the determination of optimal communication scheduling. One method includes an operation for training a machine-learning program to generate a frequency model that determines a frequency for sending communications to users. The training utilizes training data defined by features related to user information and responses of users to previous communications to the users. The method further includes determining, by the frequency model and based on information about a first user, a first frequency for the first user. The first frequency identifies the number of communications to transmit to the first user per period of time. Further, the method includes operations for receiving a communication request to send one or more communications to the first user and determining send times for the one or more communications to the first user based on the first frequency. The communications are sent at the determined send times.

Abstract: Methods, systems, and computer programs are presented for the determination of optimal communication scheduling. Send Time Optimization (STO) uses machine learning (ML) to recommend a personalized send time based on a recipient's past engagement patterns. The purpose of the ML model is to learn patterns in the data automatically and use the patterns to make personalized predictions for each recipient. The send time recommended by the model is the time at which the model believes the recipient will be most likely to engage with the message, such as clicking or opening, and use of the send time mode is expected to increase engagement from recipients. Additional customizations include communication-frequency optimization, communication-channel selection, and engagement-scoring model.

Abstract: Devices and techniques are generally described for a speech processing routing architecture. First input data representing an input request may be received. First data including a semantic interpretation of the input request may be determined. Metadata of the first input data may be determined. The metadata may identify an entity associated with the input request. In some examples, a query may be sent to a first component. The query may include the metadata. In some examples, second data that identifies a first skill associated with the entity may be received from the first component. In various examples, the first skill may be selected for processing the first input data based at least in part on the first data and the second data.

Abstract: Anonymization is the process to remove personal information from the data. Once the data is anonymized, the data may be used for creating machine-learning models without the risk of invading anyone's privacy. One of the keys to data anonymization is to make sure that the data is really anonymized so nobody could use the anonymized data to obtain private information. Different techniques for data anonymization are presented. Further, the data anonymization techniques are tested for true anonymization by comparing the results from these techniques to a random method of guessing. If the difference in the results is below a predetermined threshold margin, the data anonymization techniques are safe and ready for use.

Abstract: Methods, systems, and computer programs are presented for the determination of optimal communication scheduling. Send Time Optimization (STO) uses machine learning (ML) to recommend a personalized send time based on a recipient's past engagement patterns. The purpose of the ML model is to learn patterns in the data automatically and use the patterns to make personalized predictions for each recipient. The send time recommended by the model is the time at which the model believes the recipient will be most likely to engage with the message, such as clicking or opening, and use of the send time mode is expected to increase engagement from recipients. Additional customizations include communication-frequency optimization, communication-channel selection, and engagement-scoring model.

Abstract: Methods, systems, and computer programs are presented for the determination of optimal communication scheduling. One method includes an operation for training a machine-learning program to generate a frequency model that determines a frequency for sending communications to users. The training utilizes training data defined by features related to user information and responses of users to previous communications to the users. The method further includes determining, by the frequency model and based on information about a first user, a first frequency for the first user. The first frequency identifies the number of communications to transmit to the first user per period of time. Further, the method includes operations for receiving a communication request to send one or more communications to the first user and determining send times for the one or more communications to the first user based on the first frequency. The communications are sent at the determined send times.

Abstract: Disclosed herein is a system and method for generating an enhanced index for documents contained in a knowledge base. Documents or configuration snapshots are tokenized and then passed through a number of filters. The filters modify the token stream to generate an enhanced token stream for the document by removing words from the stream and adding domain level knowledge to the tokens in the stream. The token stream is then added to the index for the document so that searches can be completed against the knowledge base when searches are generated from configuration snapshots or from inputted search queries.

Abstract: Disclosed herein is a system and method for searching or processing queries for searching for documents contained in a domain specific knowledge base. The system takes a query and generates from the query a modified version of the query by passing the query through one or more filters in a query processor. The query processor adds or removes terms from the query. The query processor can add or recognize that two words that appear to be separate words actually identify a specific software entity or can determine that a number appearing in a query is not just a number but refers to a specific version or a number relevant to the specific problem.

Abstract: Disclosed herein is a system and method for generating an enhanced index for documents contained in a knowledge base. Documents or configuration snapshots are tokenized and then passed through a number of filters. The filters modify the token stream to generate an enhanced token stream for the document by removing words from the stream and adding domain level knowledge to the tokens in the stream. The token stream is then added to the index for the document so that searches can be completed against the knowledge base when searches are generated from configuration snapshots or from inputted search queries.

Abstract: Disclosed herein is a system and method for generating an enhanced index for documents contained in a knowledge base. Documents or configuration snapshots are tokenized and then passed through a number of filters. The filters modify the token stream to generate an enhanced token stream for the document by removing words from the stream and adding domain level knowledge to the tokens in the stream. The token stream is then added to the index for the document so that searches can be completed against the knowledge base when searches are generated from configuration snapshots or from inputted search queries.

Abstract: Disclosed herein is a system and method for taking a snapshot or input from a source and identifying appropriate documents in a knowledge base that are applicable to the input. The system identifies documents that are applicable to the query by identifying comparative features/statements found in the natural language text documents and evaluating those comparative features with the conditions of the input. When the conditions of the comparative features evaluate with the input conditions the document is considered a match. The system processes the documents through a value type filter to understand the mathematical equivalent of the comparative feature and uses this mathematical equivalent in the evaluation of the document and input.

Abstract: A system for automatically deploying a cloud includes an interface, a deployment manager, a server manager, and a deployment task executor. The interface receives cloud deployment information which includes a cloud platform type, an operating system (OS) type, at least one server at which cloud to be deployed, and a node type subordinate to the cloud platform type of the at least one server. The deployment manager determines cloud components matching the node type subordinate to the cloud platform type as cloud components to be deployed at the at least one server. The server manager pushes an OS installation file and a configuration file matching the OS type for installation at least one server. The deployment task executor deploys the determined cloud components at the at least one server.

Abstract: Disclosed herein is a system and method for taking a snapshot or input from a source and identifying appropriate documents in a knowledge base that are applicable to the input. The system identifies documents that are applicable to the query by identifying comparative features/statements found in the natural language text documents and evaluating those comparative features with the conditions of the input. When the conditions of the comparative features evaluate with the input conditions the document is considered a match. The system processes the documents through a value type filter to understand the mathematical equivalent of the comparative feature and uses this mathematical equivalent in the evaluation of the document and input.