Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating speech from text. One of the systems includes one or more computers and one or more storage devices storing instructions that when executed by one or more computers cause the one or more computers to implement: a sequence-to-sequence recurrent neural network configured to: receive a sequence of characters in a particular natural language, and process the sequence of characters to generate a spectrogram of a verbal utterance of the sequence of characters in the particular natural language; and a subsystem configured to: receive the sequence of characters in the particular natural language, and provide the sequence of characters as input to the sequence-to-sequence recurrent neural network to obtain as output the spectrogram of the verbal utterance of the sequence of characters in the particular natural language.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for speech synthesis. The methods, systems, and apparatus include actions of obtaining an audio representation of speech of a target speaker, obtaining input text for which speech is to be synthesized in a voice of the target speaker, generating a speaker vector by providing the audio representation to a speaker encoder engine that is trained to distinguish speakers from one another, generating an audio representation of the input text spoken in the voice of the target speaker by providing the input text and the speaker vector to a spectrogram generation engine that is trained using voices of reference speakers to generate audio representations, and providing the audio representation of the input text spoken in the voice of the target speaker for output.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer-readable storage media, for speech recognition using attention-based sequence-to-sequence models. In some implementations, audio data indicating acoustic characteristics of an utterance is received. A sequence of feature vectors indicative of the acoustic characteristics of the utterance is generated. The sequence of feature vectors is processed using a speech recognition model that has been trained using a loss function that uses a set of speech recognition hypothesis samples, the speech recognition model including an encoder, an attention module, and a decoder. The encoder and decoder each include one or more recurrent neural network layers. A sequence of output vectors representing distributions over a predetermined set of linguistic units is obtained. A transcription for the utterance is obtained based on the sequence of output vectors. Data indicating the transcription of the utterance is provided.

Abstract: A display panel, a display module and a manufacture method thereof, and a communication device are provided. The display panel includes a display substrate and a multiplexing circuit arranged on a first surface of the display substrate. The display panel further includes a wave-absorbing material layer arranged on a second surface of the display substrate. The wave-absorbing material layer is configured to absorb electromagnetic wave interference signals. The second surface is a surface opposite to the first surface.

Abstract: A modified nucleoside or nucleotide, the 3?-OH of the modified nucleoside or nucleotide being reversibly blocked; meanwhile, the present invention also relates to a kit comprising the nucleoside or nucleotide, and a sequencing method based on the nucleoside or nucleotide.

Abstract: The present disclosure relates to a rear sub-frame assembly, applied to an electric vehicle main body, and rear sub-frame assembly comprises a rear sub-frame main body and positioning supports, rear sub-frame main body is bilateral symmetry structure; vehicle body connecting parts connected with rear sub-frame main body on upper portion of rear sub-frame main body, and battery pack guard plate connecting parts connected with rear sub-frame main body on lower portion of rear sub-frame main body are constructed on rear sub-frame main body, stabilizing rod mounting parts, upper control arm mounting parts and lower control arm mounting parts are further arranged on rear sub-frame main body respectively; the quantity of positioning supports respectively arranged close to left end and right end of rear sub-frame main body is two, and positioning parts matched with external positioning members to position rear sub-frame main body are constructed on two positioning supports respectively.

Abstract: Systems and method are provided for tracking online communities. One method includes at a first time, sorting the plurality of users by: determining that a group of the plurality of users belongs to a community, wherein the community has a community identification of zero; labeling each user in the group with the community identification of zero; labeling each user in the group with an algorithm identification, wherein the algorithm identification is associated with the community; determining that that one user of the group is a core user; and increasing the community identification counter by one; repeat the sorting until each user of the plurality of users is labeled with a community identification and an algorithm identification; and generate a community dynamics analysis based on the sorting of the plurality of users.

Abstract: Systems and method are provided for fraud detection and user account deduplication. One method includes receiving a request from a user to register a third user account; receiving user information associated with the third user account, wherein the user information comprises a second attribute; at a third time entry, modifying the first data store by: searching the plurality of first entries in the first data store; comparing the second attribute to the first attribute of each first entry; determining that at least one first entry comprises a first attribute that is identical to the second attribute; adding second and third entries, wherein the second entry comprises the first and third user accounts, the second attribute, and the first and third time entries; and the third entry comprises the second and third user accounts, the second attribute, and the second and third time entries.

Abstract: A computer-implemented method and system for visualizing card transaction fraud analysis is presented. Transaction data and account data related to one or more payment card accounts is stored in a database. The transaction data includes a fraud score. A computer processor generates one or more of a plurality of visualizations of activity of at least one suspicious account from the one or more payment card accounts for display in a graphical user interface, each of the plurality of visualizations providing at least a graphical representation of the transaction data and which is selectable from a menu provided by the computer processor in the graphical user interface. The visualizations assist in case judgment of the one or more payment cards.

Abstract: Systems and method are provided for fraud detection and user account deduplication. One method includes receiving a request from a user to register a third user account; receiving user information associated with the third user account, wherein the user information comprises a second attribute; at a third time entry, modifying the first data store by: searching the plurality of first entries in the first data store; comparing the second attribute to the first attribute of each first entry; determining that at least one first entry comprises a first attribute that is identical to the second attribute; adding second and third entries, wherein the second entry comprises the first and third user accounts, the second attribute, and the first and third time entries; and the third entry comprises the second and third user accounts, the second attribute, and the second and third time entries.

Abstract: Methods and systems for dynamic IP categorization include receiving electronic requests to access an electronic server; logging a first set of requests occurring during a first period of time and originating from an IP address belonging to a set of IP addresses; assigning the set of IP addresses to a first category according to the first set of requests; logging a second set of requests occurring during a second period of time and originating from the first IP address or a second IP address belonging to the set of IP addresses; determining a second category according to the second set of requests; assigning the set of IP addresses to the second category when the first category and the second category differ; and providing a response to a requesting IP address based on the category associated with the set of IP addresses to which the requesting IP address belongs.

Abstract: Methods and systems for dynamic IP categorization include receiving electronic requests to access an electronic server; logging a first set of requests occurring during a first period of time and originating from an IP address belonging to a set of IP addresses; assigning the set of IP addresses to a first category according to the first set of requests; logging a second set of requests occurring during a second period of time and originating from the first IP address or a second IP address belonging to the set of IP addresses; determining a second category according to the second set of requests; assigning the set of IP addresses to the second category when the first category and the second category differ; and providing a response to a requesting IP address based on the category associated with the set of IP addresses to which the requesting IP address belongs.

Abstract: Systems and method are provided for tracking online communities. One method includes at a first time, sorting the plurality of users by: determining that a group of the plurality of users belongs to a community, wherein the community has a community identification of zero; labeling each user in the group with the community identification of zero; labeling each user in the group with an algorithm identification, wherein the algorithm identification is associated with the community; determining that that one user of the group is a core user; and increasing the community identification counter by one; repeat the sorting until each user of the plurality of users is labeled with a community identification and an algorithm identification; and generate a community dynamics analysis based on the sorting of the plurality of users.

Abstract: Systems and method are provided for fraud detection and user account deduplication. One method includes receiving a request from a user to register a third user account; receiving user information associated with the third user account, wherein the user information comprises a second attribute; at a third time entry, modifying the first data store by: searching the plurality of first entries in the first data store; comparing the second attribute to the first attribute of each first entry; determining that at least one first entry comprises a first attribute that is identical to the second attribute; adding second and third entries, wherein the second entry comprises the first and third user accounts, the second attribute, and the first and third time entries; and the third entry comprises the second and third user accounts, the second attribute, and the second and third time entries.

Abstract: Disclosed embodiments provide systems and methods related to collecting return items using an automated kiosk based on a real time risk decision. The automated kiosk captures return item information representing a return item and transmits the return item information and a request for return risk level relating to the return item to a server operable to execute a machine learning model trained on historical information to determine the risk level. The server determines the risk level based on the received return by using the machine learning model and transmits the determined risk level to the kiosk in real-time. Based on the determined risk level and a return amount associated with the return item, the server may also process a refund in real-time.

Abstract: Disclosed embodiments provide systems and methods related to collecting return items using an automated kiosk based on a real time risk decision. The automated kiosk captures return item information representing a return item and transmits the return item information and a request for return risk level relating to the return item to a server operable to execute a machine learning model trained on historical information to determine the risk level. The server determines the risk level based on the received return by using the machine learning model and transmits the determined risk level to the kiosk in real-time. Based on the determined risk level and a return amount associated with the return item, the server may also process a refund in real-time.

Abstract: A system for generating alerts including processors and storage devices. The instructions configure the one or more processors to perform operations, which include receiving an event from a data stream, extracting keys from the event, associating the event with at least one account based on the extracted keys, identifying a state variable associated with the at least one account, updating the state variable by accumulating the event in the state variable, registering a time stamp for the event in the state variable, and retiring expired events from the state variable, The operations may also include determining whether the state variable is above a threshold level and generating an alert for the account when the state variable is above the threshold level.

Abstract: Disclosed embodiments provide systems and methods related to collecting return items using an automated kiosk based on a real time risk decision. The automated kiosk captures return item information representing a return item and transmits the return item information and a request for return risk level relating to the return item to a server operable to execute a machine learning model trained on historical information to determine the risk level. The server determines the risk level based on the received return by using the machine learning model and transmits the determined risk level to the kiosk in real-time. Based on the determined risk level and a return amount associated with the return item, the server may also process a refund in real-time.

Abstract: A system for generating alerts including processors and storage devices. The instructions configure the one or more processors to perform operations, which include receiving an event from a data stream, extracting keys from the event, associating the event with at least one account based on the extracted keys, identifying a state variable associated with the at least one account, updating the state variable by accumulating the event in the state variable, registering a time stamp for the event in the state variable, and retiring expired events from the state variable. The operations may also include determining whether the state variable is above a threshold level and generating an alert for the account when the state variable is above the threshold level.