Abstract: Techniques for using a deep generative model to generate synthetic data sets that can be used to boost the performance of a discriminative model are described. In an example, an autoencoding generative adversarial network (AEGAN) is trained to generate the synthetic data sets. The AEGAN includes an autoencoding network and a generative adversarial network (GAN) that share a generator. The generator learns how to the generate synthetic data sets based on a data distribution from a latent space. Upon training the AEGAN, the generator generates the synthetic data sets. In turn, the synthetic data sets are used to train a predictive model, such as a convolutional neural network for gaze prediction.

Abstract: Systems, methods, and other embodiments described herein relate to improving the performance of a device in different geographic locations by using transfer learning to provide a customized learning model for the different locations. In one embodiment, a method includes receiving segments of a model from separate members in a geographic hierarchy and assembling the segments into the model. The segments include at least a first segment, a second segment, and a third segment. The method includes processing sensor data using the model to provide an output for assisting a device.

Abstract: A method of generating a trained trigger word detection model includes training an auxiliary model, based on an auxiliary task, to concentrate on one or more utterances and/or learn context of the one or more utterances using generic single word and/or phrase training data; and obtaining a trigger word detection model by retraining one or more final layers of the auxiliary model, which is weighted based on the auxiliary task, based on a trigger word detection task that detects one or more trigger words. The retraining uses training data specific to the one or more trigger words.

Abstract: A learning assistance device according to the disclosure is designed to perform relearning for a processing part including a learning device that has already undergone learning for performing a prescribed output from a prescribed input, the learning assistance device including: an assessment part for assessing an anomaly of the input on the basis of a prescribed reference; and a relearning part for, if the assessment part has assessed that the input is anomalous, carrying out relearning of the learner under a prescribed condition using the anomalous input as additional learning data.

Abstract: A sample image obtaining section repetitively obtains a sample image generated by photographing a given sample by a photographing unit. A feature quantity extracting section generates feature quantity data corresponding to the sample image, in reference to the sample image. The feature quantity extracting section classifies each of a plurality of pieces of feature quantity data into either training data or evaluation data. An evaluation learning section performs learning of an evaluation discriminator by using a plurality of pieces of the training data. A photographing necessity/unnecessity determining section determines necessity/unnecessity of additional photographing of the sample by using the evaluation discriminator in which the learning using the plurality of pieces of the training data has already been performed and a plurality of pieces of the evaluation data.

Abstract: The present disclosure provides a data processing method, a board card device, a computer equipment, and a storage medium for data quantization. The board card provided in the present disclosure includes a storage component, an interface device, a control component, and an artificial intelligence chip of a data processing device, where the artificial intelligence chip is connected to the storage device, the control device, and the interface apparatus, respectively. The storage component is configured to store data; the interface device is configured to implement data transmission between the artificial intelligence chip and an external equipment; and the control component is configured to monitor a state of the artificial intelligence chip.

Abstract: Mechanisms for document processing and analysis can include receiving a document and identifying, in a data structure, a record corresponding to the document. The record can include one or more entries, where each entry contains data reflecting a respective item of information extracted from a corresponding part of the document. The mechanisms can include determining for each entry of the record, a corresponding degree of association between the entry and a respective item of information referenced by the entry. They can further include updating the corresponding degrees of association, and selecting, among the corresponding degrees of association, a set of corresponding degrees of association whose aggregate degree of association satisfies a criterion.

Abstract: Systems and methods are provided that include a processor executing a program to receive input text from a user. The processor is further configured to, for a predetermined number of iterations, input an initial image into a diffusion process to generate a processed image, back-propagate the processed image through a text-image match gradient calculator to calculate a gradient against the input text, and update the initial image with an image generated by applying the calculated gradient to the processed image. The pixel values of the processed image during a first portion of the predetermined number of iterations are value clamped to a first range, and pixel values of the processed image during a second portion of the predetermined number of iterations are value clamped to a second range that is a subset of the first range.

Abstract: Devices, systems, and methods are provided for brand matching using multi-layer machine learning. A method may include generating, based on a first embedding vector and a second embedding vector as inputs to a twin neural network, a third embedding vector and a fourth embedding vector; generating, based on the first embedding vector and the second embedding vector as inputs to a difference neural network, a difference vector indicative of a difference between the first embedding vector and the second embedding vector; generating a concatenated vector by concatenating the third embedding vector with the fourth embedding vector and the difference vector; generating, based on the concatenated vector as an input to a feedforward neural network (FFN), a score between zero and one, the score indicative of a relationship between a first entity and a second entity.

Abstract: The article pickup system includes: a camera that is provided in a mobile body capable of moving in an environment where a plurality of articles to be picked up are present and that acquires image data by capturing the plurality of articles; and an article pickup robot that moves in the environment and picks up the plurality of captured articles. The article pickup robot includes: a camera position acquisition unit that detects the camera and then acquires position information of the camera; an article position specification unit that specifies positions of the plurality of articles based on the position information of the camera; a mobile unit that moves to the specified positions; and a pickup unit that picks up the plurality of articles at the positions to which the mobile unit has moved.

Abstract: A warehouse storage and retrieval system including an array of multilevel storage racks having at least one transfer deck, picking isles and storage areas disposed along picking isles, the storage areas being configured to hold differing loads, and a controller including a management module configured to variably size the storage areas of the array of multilevel storage rack modules and assign each of the variably sized storage areas to a corresponding one of the differing loads, wherein the storage and retrieval system is arranged to transport the differing loads for placement in the variably sized storage areas assigned by the controller.

Abstract: Various aspects described herein relate to a system that utilized deep learning and neural networks to estimate/predict an amount of natural resource production in a well given a set of parameters indicative of physical changes to the well. In one aspect, a virtual flow meter includes memory having computer-readable instructions stored therein and one or more processors configured to execute the computer-readable instructions to receive one or more input parameters indicative of physical changes to at least one well; apply the one or more input parameters to a trained neural network architecture; and determine one or more outputs of the trained neural network architecture, the one or more outputs corresponding to predicted fluid output of the at least one well.

Abstract: A method includes receiving text input data for conversion into synthesized speech and determining, using a hotword-aware model trained to detect a presence of a hotword assigned to a user device, whether a pronunciation of the text input data includes the hotword. The hotword is configured to initiate a wake-up process on the user device for processing the hotword and/or one or more other terms following the hotword in the audio input data. When the pronunciation of the text input data includes the hotword, the method also includes generating an audio output signal from the text input data and providing the audio output signal to an audio output device to output the audio output signal. The audio output signal when captured by an audio capture device of the user device, configured to prevent initiation of the wake-up process on the user device.

Abstract: A damage estimation device includes: an operation parameter reception unit that acquires an operation parameter related to an operation of a work machine; a damage estimation model storage unit that stores a damage estimation model constructed by machine learning using training data with the operation parameter as an input value and a damage parameter related to damage in a predetermined portion of the work machine as an output value; and a damage parameter estimation unit that estimates the damage parameter by inputting the operation parameter acquired by the operation parameter reception unit to the damage estimation model stored in the damage estimation model storage unit.

Abstract: Computer systems and methods train a deep neural network through machine learning. In response to detection of a training condition, computer system replaces a target node of the network with a split detector compound node, where, prior to replacement, the target node detected a pattern that activated the target node beyond a specified threshold. The split detector compound node comprises first and second nodes, such that: the first node is activated when significant evidence exists in favor of detection of the pattern in inputs to the first node; and the second node is activated when significant evidence exists against detection of the pattern in inputs to the second node, such that activations of the first and second nodes are computed independently. After replacing the target node with the split detector compound node, training of the network through machine learning is resumed.

Abstract: A wireless network can generate candidate signal configurations for physical transmissions to or from a user equipment (UE) in a radio environment. The generation of candidate signal configurations can be performed using a first neural network that is associated with the UE. These signal configurations can then be evaluated using a second neural network that is associated with the radio environment. The second neural network can be trained using measurements from previous physical transmissions in the radio environment. The trained second neural network generates a reward value that is associated with the candidate signal configurations. The first neural network is then trained using the reward values from the second neural network to produce improved candidate signal configurations. When a signal configuration that produces a suitable reward value is generated, this signal configuration can be used for the physical transmission in the radio environment.

Abstract: The present invention relates to a cloud simulation apparatus and method for verifying a rail transit-oriented full-automatic unmanned driving scene, where the apparatus includes a cloud access terminal and a cloud server terminal. The cloud server terminal includes a central dispatching module, a station control module, a rail-mounted module, an interface logic management layer, and a device base layer, the cloud access terminal is connected to the central dispatching module, the central dispatching module is connected to the station control module and the rail-mounted module respectively, every two of the station control module, the rail-mounted module, and the interface logic management layer are connected, and the interface logic management layer is connected to the device base layer.

Abstract: A device may receive demand data associated with a product or a service, multiple forecasting models, and multiple cost functions, and may identify primary parameters for the multiple models based on the demand data. The device may utilize a model to rank the multiple forecasting models and the multiple cost functions based on the primary parameters, and may select optimum primary parameters based on ranking the multiple forecasting models and the multiple cost functions. The device may identify secondary parameters for the multiple forecasting models based on the demand data and the optimum primary parameters. The device may select optimum secondary parameters based on ranking the multiple forecasting models and the multiple cost functions, and may select a forecasting model, from the multiple forecasting models, based on the optimum primary parameters and the optimum secondary parameters. The device may perform one or more actions based on selecting the forecasting model.

Abstract: After inputting input data to a floating pre-trained convolution neural network to generate floating feature maps for each layer of the floating pre-trained CNN model, a statistical analysis on the floating feature maps is performed to generate a dynamic quantization range for each layer of the floating pre-trained CNN model. Based on the obtained quantization range for each layer, the proposed quantization methodologies quantize the floating pre-trained CNN model to generate the scalar factor of each layer and the fractional bit-width of a quantized CNN model. It enables the inference engine performs low-precision fixed-point arithmetic operations to generate a fixed-point inferred CNN model.

Abstract: A method for determining a stereotactic brain target includes at least one target point, the method comprising the following steps: selecting patients for whom the result measured following treatment at at least one target point is greater than or equal to a threshold, postoperative imaging having been performed for each of the patients; processing the postoperative imaging so as to determine coordinates of the at least one target point; selecting brain marker points; processing the postoperative imaging so as to determine coordinates of the marker points; creating a learning database comprising the coordinates of the target points and the coordinates of the marker points determined for all of the selected patients; determining a prediction function giving the coordinates of at least one target point according to the coordinates of the marker points by using the learning database and a supervised statistical learning method; processing preoperative imaging of a new patient to be treated so as to determine coo

Abstract: Aspects for backpropagation of a multilayer neural network (MNN) in a neural network processor are described herein. The aspects may include a computation module configured to receive one or more groups of MNN data. The computation module may further include a master computation module configured to calculate an input gradient vector based on a first output gradient vector from an adjacent layer and based on a data type of each of the one or more groups of MNN data. Further still, the computation module may include one or more slave computation modules configured to parallelly calculate portions of a second output vector based on the input gradient vector calculated by the master computation module and based on the data type of each of the one or more groups of MNN data.

Abstract: The information processing apparatus obtains a distribution characteristic representing an attribute of a color included in an acquired first image group and determines the number of times of learning based on the distribution characteristic. The information processing apparatus generates a data set consisting of a set of the first image group and a second image group corresponding to the first image group and by using the generated data set, generates a learning model by performing learning using a network based on the determined number of times of learning.

Abstract: A method and apparatus for improving the quality of an attention-based sequence-to-sequence model. The method includes determining an output sequence corresponding to an input sequence based on an attention-based sequence-to-sequence model, selecting at least one target attention head from among a plurality of attention heads, detecting at least one error output token among output tokens constituting the output sequence based on the target attention head, and correcting the output sequence based on the error output token.

Abstract: One embodiment of the present disclosure provides a system and method for determining a decision-making strategy. During operation, the system can obtain sample data corresponding to a service. The system can pre-determine, based on the sample data and the service, a basic parametric shape model and can generate a plurality of shape parameters for the basic parametric shape mode. The system can then generate one or more parametric shape models with different shapes based on the plurality of shape parameters. The system can determine goodness-of-fit of the one or more parametric shape models with respect to a pre-defined decision-making strategy. Next, the system can in response to determining that the goodness-of-fit of at least one parametric shape model satisfies a set of objectives, outputting the at least one parametric shape model as an optimum decision-making strategy for the service.

Abstract: This disclosure relates to a method of generating a symbol string based on an input sentence represented by a sequence of symbols. In particular, the method involves receiving an input symbol string representing a sentence, generating, using a neural network based on a sequence of dependency structure of elements in the input symbol string, an output symbol string corresponding to the input sentence. The neural network includes an encoder that converts elements of the input symbol string to a first hidden state in a form of a multi-dimensional vector, an attention mechanism that applies a weight to the first hidden state and generates the weighted first hidden state as a second hidden state, a decoder that generates a third hidden state based on at least one element of the input symbol string, an element of the output symbol string, and the second hidden state, and an output generator that generates an element of the output symbol string based on the second hidden state and the third hidden state.

Abstract: This disclosure involves verifying that a power network model corresponds to an electric power network providing electrical power in a geographical area. For instance, a validation device computes a validation score for a power network model based on a connectivity score, an asset score, and a power-flow score. The connectivity score indicates connectivity errors in the power network model as compared to the power network. The asset score indicates power-delivery errors in the power network model with respect to power-consuming assets serviced by the power network. The power-flow score indicates power-flow calculation errors in the power network model with respect to voltage ranges for the power network. The validation score is repeatedly computed for iteratively updated versions of the power network model until a threshold validation score is obtained. The validated power network model is provided to a control system for identifying and correcting errors in the power network.

Abstract: Systems, methods and computer software are disclosed for providing an OpenRAN solution suite. In one embodiment, a method is disclosed, the method including communicating, by an all G COTS (Commercial off the Shelf) Base Band Unit (BBU), with a plurality of different G user devices; communicating, by a software platform, with the all G COTS BBU, wherein the software platform includes virtualized software providing open RAN controller functionality, network orchestrator functionality, and SON edge core functionality; and communicating, by the software platform, with a plurality of different G core networks.

Abstract: A method including receiving, in a machine learning model (MLM), a corpus including words. The MLM includes layers configured to extract keywords from the corpus, plus a retrospective layer. A first keyword and a second keyword from the corpus are identified in the layers. The first and second keywords are assigned first and second probabilities. Each probability is a likelihood that a keyword is to be included in a key phrase. A determination is made, in the retrospective layer, of a first probability modifier that modifies the first probability based on a first dependence relationship between the second keyword being placed after the first keyword. The first probability is modified using the first probability modifier. The first modified probability is used to determine whether the first keyword and the second keyword together form the key phrase. The key phrase is stored in a non-transitory computer readable storage medium.

Abstract: A method and an apparatus for creating a workflow, the method being performed by a computing device, and including obtaining a log file generated while a user performs a task, parsing each of a plurality of events recorded in the log file, and creating an activity sequence to correspond to an order of the events, grouping a plurality of partial sequences extracted from the activity sequence, and creating a plurality of activity groups to correspond to each of a plurality of different repetitive tasks included in the task and creating the workflow of at least one activity group among the plurality of activity groups.

Abstract: A system and a method are disclosed that generate for display to a remote operator a user interface comprising a map, the map comprising visual representations of a source area, a plurality of candidate robots, and a plurality of candidate destination areas. The system receives, via the user interface, a selection of a visual representation of a candidate robot of the plurality of candidate robots, and detects a drag-and-drop gesture within the user interface of the visual representation of the candidate robot being dragged-and-dropped to a visual representation of a candidate destination area of the plurality of candidate destination areas. Responsive to detecting the drag-and-drop gesture, the system generates a mission, where the mission causes the candidate robot to autonomously transport an object from the source area to the candidate destination area.

Abstract: The present invention relates to an integrated sleep diagnosis and treatment device, and more particularly to an integrated apnea diagnosis and treatment device. The present invention additionally relates to methods of sleep diagnosis and treatment. The sleep disorder treatment system of the present invention can use a diagnosis device to perform various forms of analysis to determine or diagnose a subject's sleeping disorder or symptoms of a subject's sleep disorder, and using this analysis or diagnosis can with or in some embodiments without human intervention treat the subject either physically or chemically to improve the sleeping disorder or the symptoms of the sleeping disorder. The diagnostic part of the system can use many different types of sensors and methods for diagnosing the severity of the symptoms of or the sleep disorder itself. The treatment part of the system can use a device to physically or chemically treat the subject's symptoms or sleep disorder itself.

Abstract: Systems and methods for simultaneously presenting to a user with a plurality of available options for resolving delinquent debts to local, state and national tax authorities and providing the user with appropriate documentation to resolve his/her delinquency status based on a selected option. The system uses all available information from the tax agency website and personal information of the taxpayer, such as expenses and income, to compute the most favorable resolution solution for the taxpayer. The system further uses deep learning to ranks the generated options based on previously accepted solutions by the tax agency.

Abstract: The present disclosure relates to a pre-5th-Generation (5G) or 5G communication system to be provided for supporting higher data rates Beyond 4th-Generation (4G) communication system such as Long Term Evolution (LTE). The embodiments of the present invention provide methods and devices for network access. The method includes receiving a first request message forwarded by a distributed unit in a base station and indication information of the distributed unit with respect to the first request message, the first request message requesting to connect a user equipment to a network; determining a processing to be performed on the first request message based on the indication Information; and transmitting an indication of the determined processing to the distributed unit.

Abstract: The invention relates to a method for testing an air traffic management electronic system. The method includes receiving by the system of input data representative of a state of air traffic and establishing by the system of information related to air traffic as a function of the input data and delivery of the information to an electronic test device of the system. The method further includes determining by the electronic test device, as a function of the delivered information, of air traffic control instructions and providing said system with the instructions and receiving and processing of the instructions by the system.

Abstract: In various example embodiments, a system and method for data mesh visualization are presented. An avatar representation of a user is generated. A user-specified change to the avatar representation is received and the avatar representation is changed based on the user-specified change. A current status of the user is determined, and an updated avatar representation is generated based on the current status of the user. The updated avatar is displayed in association with a communication of the user.

Abstract: Methods include training, using a generative adversarial network, a generator model using data noise that includes data corresponding to real problems of a telecommunication site, generating a generated problem that has not occurred at the telecommunication site and that has a non-zero probability of occurring at the site in the future, providing the generated problem to a virtual agent that is configured to generate a solution action to resolve the generated problem, evaluating the solution action relative to the generated problem to determine a performance value corresponding to the solution action, and responsive to the performance value being higher than other performance values corresponding to other solution actions for the generated problem, generating a generic problem model that corresponds to the generated problem and that is associated with the solution action.

Abstract: Disclosed is a non-transitory computer readable medium storing a computer program, wherein the computer program includes instructions to perform following steps for data processing when the computer program is executed by one or more processors, the steps including: recognizing at least one continuous section from each raw data subset; determining at least one serialization point, based on a start point and an end point of each of the at least one continuous section for each of the raw data subset; and generating a training data set by generating serialized training data, based on the at least one serialization point.

Abstract: A storage system configured to house a plurality of containers housing inventory items includes support members, a first set of parallel rails to support a mobile, manipulator robot, and a fluid supply line having a plurality of valves disposed within the fluid supply line. Each of the valves having a closed condition in which the supply line is in fluid isolation from an outside environment and an open condition in which the supply line is in fluid communication with the environment such that the supply line is configured to supply fluid to a mobile, manipulator robot. Mobile, manipulator robots for retrieving inventory items stored within the containers and retrieval methods are also disclosed herein.

Abstract: Computer systems and methods modify a base deep neural network (DNN). The method comprises replacing the target node of the base DNN with a compound node to thereby create a modified base DNN. The compound node comprises at least first and second nodes. The first node is trained to detect target node patterns in inputs to the first node and the second node is trained to detect an absence of the target node patterns in inputs to the second node, and the first and second nodes are trained to be non-complementary.

Abstract: One or more embodiments herein can facilitate a process to guide a user at a digital medium. An exemplary system can comprise a memory that stores computer executable components, and a processor that executes the computer executable components stored in the memory, wherein the computer executable components can comprise an obtaining component that obtains user behavior data from a digital medium, the user behavior data defining an interaction of the user with the digital medium, a prediction component that, based on the user behavior data, predicts a path of the user within the digital medium, and a classification component that, based on the user behavior data and on the path prediction, classifies a position of the user along a defined path to a conversion objective of the digital medium. The conversion objective can comprise obtaining the user as a customer, achieving a financial transaction, or presentation of a communication.

Abstract: The disclosed computer-implemented method for utilizing federated machine-learning to protect against potentially malicious data may include (i) arranging a set of client devices into groups for applying a federated machine-learning model, (ii) determining model updates for each of the groups over a predetermined period, (iii) training one or more recurrent neural networks to derive a low-dimensional representation of the model updates, (iv) calculating a data quality score for each of the client devices based on the model updates, (v) applying the federated machine-learning model to classify data instances on each of the client devices as including clean data or potentially corrupt data, and (vi) performing a security action that protects against the potentially malicious data by tagging the data instances classified as the potentially corrupt data. Various other methods, systems, and computer-readable media are also disclosed.

Abstract: Techniques for automatic scene change detection are described. As one example, a computer-implemented method includes receiving a request to train an ensemble of machine learning models on a training dataset of videos having labels that indicate scene changes to detect a scene change in a video, partitioning each video file of the training dataset of videos into a plurality of shots, training the ensemble of machine learning models into a trained ensemble of machine learning models based at least in part on the plurality of shots of the training dataset of videos and the labels that indicate scene changes, receiving an inference request for an input video, partitioning the input video into a plurality of shots, generating, by the trained ensemble of machine learning models, an inference of one or more scene changes in the input video based at least in part on the plurality of shots of the input video, and transmitting the inference to a client application or to a storage location.

Abstract: The present disclosure discloses an artificial intelligence apparatus including an input interface configured to obtain input data, a sensing interface configured to obtain environment information, and one or more processors configured to classify an object by inputting the input data obtained from the input interface to an artificial intelligence model, in which the artificial intelligence model uses a first learning model and a second learning model which is connected with the first learning model and includes a plurality of output layers to respectively assign weights to the respective result values output by the plurality of output layers and combine the respective result values to which the weights are assigned to derive the final result.

Abstract: Conventional email filtering services are not suitable for recognizing sophisticated malicious emails, and therefore may allow sophisticated malicious emails to reach inboxes by mistake. Introduced here are threat detection platforms designed to take an integrative approach to detecting security threats. For example, after receiving input indicative of an approval from an individual to access past email received by employees of an enterprise, a threat detection platform can download past emails to build a machine learning (ML) model that understands the norms of communication with internal contacts (e.g., other employees) and/or external contacts (e.g., vendors). By applying the ML model to incoming email, the threat detection platform can identify security threats in real time in a targeted manner.

Abstract: Embodiments of the present disclosure are directed to a system, methods, and computer-readable media for facilitating stylistic expression transfers in machine translation of source sequence data. Using integrated loss functions for style transfer along with content preservation and/or cross entropy, source sequence data is processed by an autoencoder trained to reduce loss values across the loss functions at each time step encoded for the source sequence data. The target sequence data generated by the autoencoder therefore exhibits reduced loss values for the integrated loss functions at each time step, thereby improving content preservation and providing for stylistic expression transfer.

Abstract: A polarizing beam splitting system is described. The polarizing beam splitting system may include first and second prisms where the volume of the first prism is no greater than half the volume of the second prism. The first prism includes first and second surfaces and a light source may be disposed adjacent the first surface and an image forming device may be disposed adjacent the second surface. The first prism has a first hypotenuse and the second prism has a second hypotenuse. A reflective polarizer is disposed between the first and second hypotenuses.

Abstract: A method may include obtaining data representative of an NNT of a graph-based model that includes multiple components. The multiple components may include multiple neural nodes and at least one connection. The at least one connection may associate two or more of the neural nodes. The method may include displaying the NNT including the multiple components in a GUI via a display screen. The method may include receiving user input effective to indicate that at least one of the components of the NNT is to be modified. The user input may be received via the GUI. The method may include modifying the at least one of the components of the NNT. The at least one of the components may be modified based on the user input. The method may include displaying the NNT that comprises the modification of the at least one of the components via the GUI.

Abstract: A hearing assistance device including: a microphone arranged to receive sound of an environment in which the hearing assistance device is located; a wireless communication device arranged to wirelessly communicate data with an external device; a controller operably connected with the microphone and arranged to process the received sound using sound processing settings that have been determined using a trained machine learning processing model; and a speaker operably connected with the controller and arranged to output the processed sound. The sound processing settings have been determined using the trained machine learning processing model based on a hearing response of a user and one or more properties of the environment.

Abstract: A first and second blending profile may be created for a set of question answering pipelines. A set of test answer data may be generated for a first answering pipeline. The test answer data may be generated based on a set of test question and using an answer key associated with the test questions. Based on the test answer data, a first blending profile can be created for the first answering pipeline. Using the set of test questions and a second answer key, another set of test answer data may be generated. This set may be generated for the second answering pipeline. Using this second answering pipeline test answer data, a second blending profile can be generated for the second answering pipeline. Each blending profile may have metadata about a confidence of each pipeline.

Abstract: A method includes obtaining (i) device data associated with multiple electronic devices and (ii) advertisement data associated with one or more advertisement campaigns. The method also includes identifying first features and second features, where the first features correspond to the device data and the second features correspond to the advertisement data. The method further includes generating a graph relating usage history of the multiple electronic devices and one or more advertisement genres. The method also includes identifying a specified electronic device from among the multiple electronic devices and an advertisement segment from among the one or more advertisement campaigns using the first features, the second features, and the graph. In addition, the method includes providing the advertisement segment to the specified electronic device.