Abstract: The disclosure relates to systems and methods of improved attention for neural networks. A system may access a first array of entries and a second array of entries, wherein the first array represents a first word embedding generated from one or more first characters of an input and the second array represents a second word embedding from the one or more second characters of the input. A system may generate at least a first difference matrix based on the first array and at least a second difference matrix based on the second array. A system may determine a difference value based on vector based at least in part on the difference value. A system may provide the context vector to subsequent layers of a neural network.

Abstract: Custom communication alert techniques are described. In one or more implementations, a triggering condition is detected by one or more computing devices that is found by searching data using one or more extraction rules of a late-binding schema. Responsive to the detection of the triggering condition of the alert, a communication is formed by the one or more computing devices that corresponds to the alert and that includes one or more tokens based on one or more values of the data taken from fields defined by the one or more extraction rules. The communication is caused to be transmitted by the one or more computing device via a network for receipt by at least one computing device of an intended recipient of the communication.

Abstract: The disclosure relates to systems and methods of generating a mixture model for approximating non-normal distributions of time series data. The mixture model may include clusters of normal distributions that together approximate a non-normal distribution. The mixture model may be used to normalize input data for machine learning models. For example, a machine learning model such as an autoencoder may be trained to make predictions on the normalized input data. The predictions may relate to the time series of data. In one example, the time series of data may be market data for a security. The market data my include one or more features that are normalized using the mixture model. The predictions may include a predicted rate at which a lender will charge to borrow a security for short selling, where such rate may depend on the market data for the security.

Abstract: The disclosure relates to systems and methods of generating a mixture model for approximating non-normal distributions of time series data. The mixture model may include clusters of normal distributions that together approximate a non-normal distribution. The mixture model may be used to normalize input data for machine learning models. For example, a machine learning model such as an autoencoder may be trained to make predictions on the normalized input data. The predictions may relate to the time series of data. In one example, the time series of data may be market data for a security. The market data my include one or more features that are normalized using the mixture model. The predictions may include a predicted rate at which a lender will charge to borrow a security for short selling, where such rate may depend on the market data for the security.

Abstract: Systems and methods are disclosed for scalable bucket merging in a data intake and query system. Various components of a bucket manager can be used to monitor recently-created buckets of data in common storage that are associated with a particular tenant and a particular index, apply a comprehensive bucket merge policy to determine groups of buckets that qualify for merging, merge those group of buckets into merged buckets to be stored in the common storage, and update any information associated with the merged buckets and pre-merged buckets. These components may be shared across multiple tenants, and some of these components may be dynamically scalable based on need. This approach may also provide many additional benefits, including improved search performance from merged buckets, efficient resource utilization associated with discriminate merging, and redundancy in case of component failure.

Abstract: A base station device may include a memory configured to store instructions and a processor configured to execute the instructions to obtain one or more radio signal quality parameter values associated with a user equipment (UE) device and determine a channel quality class associated with the UE device based on the obtained one or more radio signal quality parameter values using a machine learning model. The processor may be further configured to identify an application data stream associated with the UE device, select an application bandwidth for the application data stream based on the determined channel quality class, and send application data associated with the application data stream to the UE device based on the selected application bandwidth.

Abstract: The present disclosure relates to predicting non-linear subject behavior that occurs in response to stimulus content. Determining a behavior model for the subject is described. The behavior model describes a density of an observed behavior of a subject. A prior probability subject behavior distribution associated with the observed behavior is determined. The prior probability subject behavior distribution comprises an assumption describing the observed behavior. The prior probability subject behavior distribution comprises a Gamma prior probability distribution. The non-linear subject behavior is predicted based on the stimulus content, the Gamma prior probability distribution, and the behavior model.

Abstract: A system described herein may use a Bloom filter to reduce unnecessary accesses to memory. The filter may provide constant search time through hashing and use a relatively small amount of space. The filter may utilize hashing functions to generate a key associated with each data item of a set of data items, where the bit length of the key may be calibrated and tuned for selection of a minimum key length (in number of bits) versus a desired false positive rate. Such keys may be combined to form the filter. The resulting filter may be compared to various request keys that are generated using the same hashing functions in order to determine whether a requested item is present in the associated memory. The keys may be compared to the filter by comparing each bit in the key to each bit of the filter at the same location.

Abstract: Systems and methods are disclosed for scalable bucket merging in a data intake and query system. Various components of a bucket manager can be used to monitor recently-created buckets of data in common storage that are associated with a particular tenant and a particular index, apply a comprehensive bucket merge policy to determine groups of buckets that qualify for merging, merge those group of buckets into merged buckets to be stored in the common storage, and update any information associated with the merged buckets and pre-merged buckets. These components may be shared across multiple tenants, and some of these components may be dynamically scalable based on need. This approach may also provide many additional benefits, including improved search performance from merged buckets, efficient resource utilization associated with discriminate merging, and redundancy in case of component failure.

Abstract: Custom communication alert techniques are described. A triggering condition is detected by one or more computing devices that is found by searching data using one or more extraction rules of a late-binding schema. Responsive to the detection of the triggering condition of the alert, a communication is formed by the one or more computing devices that corresponds to the alert and that includes one or more tokens based on one or more values of the data taken from fields defined by the one or more extraction rules. The communication is caused to be transmitted by the one or more computing device via a network for receipt by at least one computing device of an intended recipient of the communication.

Abstract: Systems and methods are disclosed for scalable bucket merging in a data intake and query system. Various components of a bucket manager can be used to monitor recently-created buckets of data in common storage that are associated with a particular tenant and a particular index, apply a comprehensive bucket merge policy to determine groups of buckets that qualify for merging, merge those group of buckets into merged buckets to be stored in the common storage, and update any information associated with the merged buckets and pre-merged buckets. These components may be shared across multiple tenants, and some of these components may be dynamically scalable based on need. This approach may also provide many additional benefits, including improved search performance from merged buckets, efficient resource utilization associated with discriminate merging, and redundancy in case of component failure.

Abstract: A system described herein may use a Bloom filter to reduce unnecessary accesses to memory. The filter may provide constant search time through hashing and use a relatively small amount of space. The filter may utilize hashing functions to generate a key associated with each data item of a set of data items, where the bit length of the key may be calibrated and tuned for selection of a minimum key length (in number of bits) versus a desired false positive rate. Such keys may be combined to form the filter. The resulting filter may be compared to various request keys that are generated using the same hashing functions in order to determine whether a requested item is present in the associated memory. The keys may be compared to the filter by comparing each bit in the key to each bit of the filter at the same location.

Abstract: A base station device may include a memory configured to store instructions and a processor configured to execute the instructions to obtain one or more radio signal quality parameter values associated with a user equipment (UE) device and determine a channel quality class associated with the UE device based on the obtained one or more radio signal quality parameter values using a machine learning model. The processor may be further configured to identify an application data stream associated with the UE device, select an application bandwidth for the application data stream based on the determined channel quality class, and send application data associated with the application data stream to the UE device based on the selected application bandwidth.

Abstract: A method, a device, and a non-transitory storage medium that stores instructions to store one or multiple instances of first data that indicate one or multiple types of forward error correction (FEC) coding schemes supported by an end device; generate a radio resource control connection request; select one of the one or multiple instances of first data during the generation of the radio resource control connection request; include the one of the one or multiple instances of first data in the radio resource control connection request in response to the selection; and transmit the radio resource control connection request to a wireless station of a wireless network in response to the inclusion.

Abstract: Compatibility and/or compliance testing for a wireless cellular network may be performed using a testing system that includes a device that implements multiple base stations in a single hardware device. Additionally, a network simulation server may simulate a core portion of the wireless cellular network. Test cases, such as test cases defined using a scripting language, may be received and the test cases may be interpreted to obtain configuration information for the testing system. Based on the configuration information, the base stations and the simulated network devices may be initially configured. After configuration, and based on the test case, a UE may be controller to interact with the simulated network in a manner implements the desired compatibility and/or compliance tests.

Abstract: Custom communication alert techniques are described. In one or more implementations, a triggering condition is detected by one or more computing devices that is found by searching data using one or more extraction rules of a late-binding schema. Responsive to the detection of the triggering condition of the alert, a communication is formed by the one or more computing devices that corresponds to the alert and that includes one or more tokens based on one or more values of the data taken from fields defined by the one or more extraction rules. The communication is caused to be transmitted by the one or more computing device via a network for receipt by at least one computing device of an intended recipient of the communication.

Abstract: Custom communication alert techniques are described where a triggering condition is detected by one or more computing devices that is found by searching data using one or more extraction rules of a late-binding schema. Responsive to the detection of the triggering condition of the alert, a communication is formed by the one or more computing devices that corresponds to the alert and that includes one or more tokens based on one or more values of the data taken from fields defined by the one or more extraction rules. The communication is caused to be transmitted by the one or more computing device via a network for receipt by at least one computing device of an intended recipient of the communication.

Abstract: A method, a device, and a non-transitory storage medium that stores instructions to store one or multiple instances of first data that indicate one or multiple types of forward error correction (FEC) coding schemes supported by an end device; generate a radio resource control connection request; select one of the one or multiple instances of first data during the generation of the radio resource control connection request; include the one of the one or multiple instances of first data in the radio resource control connection request in response to the selection; and transmit the radio resource control connection request to a wireless station of a wireless network in response to the inclusion.

Abstract: A method, a device, and a non-transitory storage medium that stores instructions to store one or multiple instances of first data that indicate one or multiple types of forward error correction (FEC) coding schemes supported by an end device; generate a radio resource control connection request; select one of the one or multiple instances of first data during the generation of the radio resource control connection request; include the one of the one or multiple instances of first data in the radio resource control connection request in response to the selection; and transmit the radio resource control connection request to a wireless station of a wireless network in response to the inclusion.

Abstract: Techniques are described that provide for the collaborative co-existence of LTE-U base stations and access points associated with technologies, such as WiFi, that use unlicensed frequency spectrum. In one implementation, a LTE-U device may select a frequency block, in an unlicensed frequency spectrum, based on measured signal strengths and decoded Master Information Blocks (MIBs) that are broadcast in the unlicensed frequency spectrum. The LTE-U device may initiate communications via the selected frequency block by transmitting a MIB that includes a token identifier field that identifies the selected frequency block.