Abstract: Using a simple cue to reduce a number of sequential frames included in a video that needs to be analyzed by an artificial neural network to predict information corresponding to a projectile depicted within the video is provided. A timing of the simple cue associated with the video is detected. The number of sequential frames within the video is reduced down to only those frames that are within a specified range of the simple cue. The artificial neural network is used to analyze the reduced number of sequential frames. The information corresponding to the projectile is predicted based on analyzing the reduced number of sequential frames using the artificial neural network.

Abstract: One or more computer processors create a user-event localization model for an identified remote audience member in a plurality of identified remote audience members for an event. The one or more computer processors generate a virtual audience member based the identified remote audience member utilizing a trained generated adversarial network and one or more user preferences. The one or more computer processors present the generated virtual audience member in a location associated with the event. The one or more computer processors dynamically adjust a presented virtual audience member responsive to one or more event occurrences utilizing the created user-event localization model.

Abstract: One or more computer processors create a user-event localization model for an identified remote audience member in a plurality of identified remote audience members for an event. The one or more computer processors generate a virtual audience member based the identified remote audience member utilizing a trained generated adversarial network and one or more user preferences. The one or more computer processors present the generated virtual audience member in a location associated with the event. The one or more computer processors dynamically adjust a presented virtual audience member responsive to one or more event occurrences utilizing the created user-event localization model.

Abstract: Computer-implemented method includes developing, via a processor, a words model from a plurality of natural language text based articles relating to a subject and generating, via the processor, a static vector based upon the words model. The computer-implemented method further includes developing, via the processor, an actual articles model from actual articles, generating, via the processor, a bootstrapped vector using the actual articles model, generating, via the processor, a n-dimensional depth item using the static vector and the bootstrapped vector, and determining, via the processor, evidence based on the n-dimensional depth item. The computer-implemented method still further includes presenting, via the processor and a display, the evidence base upon an input query from a user.

Abstract: Event data and event participant data corresponding to a time during an event are encoded into a multidimensional feature vector using a trained encoder network. Using an attention mask, the multidimensional feature vector is adjusted, the adjusting amplifying a portion of the multidimensional feature vector according to an importance level of the portion. The adjusted multidimensional feature vector is decoded into an excitement level score using a trained decoder network. Using the excitement level score and a trained neural network model, a frequency and an amplitude of simulated crowd noise corresponding to the time during the event are generated.

Abstract: Methods, computer program products, and systems are presented. The method computer program products, and systems can include, for instance: running a comment examining process for processing comments of one or more user to provide a comment processing output; applying data of a dataset as training data for training a neural network to define a trained neural network, wherein the training data includes input node training data and output node training data; and generating a decision rule for the dataset based on a transfer function of the trained neural network, wherein the decision rule is based on the comment processing output.

Abstract: The disclosure includes using dilation of speech content from an interlaced audio input for speech recognition. A learning model is initiated to determine dilation parameters for each of a plurality of audible sounds of speech content from a plurality of speakers received at a computer as an audio input. As part of the learning model, a change of each of a plurality of independent sounds is determined in response to an audio stimulus, the independent sounds being derived from the audio input. The disclosure applies the dilation parameters, respectively, based on the change of each of the independent sounds. A voice print is constructed for each of the speakers based on the independent sounds and the dilation parameters, respectively. Speech content is attributed to each of the plurality of speakers based at least in part on the voice print, respectively, and the independent sounds.

Abstract: The disclosure includes using dilation of speech content from a separated audio input for speech recognition. An audio input from a speaker and predicted changes for the audio input based on an external noise are received at a CNN (Convolutional Neural Network). In the CNN, diarization is applied to the audio input to predict how a dilation of speech content from the speaker changes the audio input to generate a CNN output. A resulting dilation is determined from the CNN output. A word error rate is determined for the dilated CNN output to determine an accuracy for speech to text outputs. An adjustment parameter is set to change a range of the dilation based on the word error rate, and the resulting dilation of the CNN output is adjusted based on the adjustment parameter to reduce the word error rate.

Abstract: A model-pair is selected to recognize spoken words in a speech signal generated from a speech, which includes an acoustic model and a language model. A degree of disjointedness between the acoustic model and the language model is computed relative to the speech by comparing a first recognition output produced from the acoustic model and a second recognition output produced from the language model. When the acoustic model incorrectly recognizes a portion of the speech signal as a first word and the language model correctly recognizes the portion of the speech signal as a second word, a textual representation of the second word is determined and associated with a set of sound descriptors to generate a training speech pattern. Using the training speech pattern, the acoustic model is trained to recognize the portion of the speech signal as the second word.

Abstract: A relevant factoid(s) related to multimedia data is generated by splitting a multimedia item into a media component and a text component. Text information is retrieved relevant to text data from the text component using a query. The text information is summarized into a factoid. Source data is checked for an image based on the multimedia component. A current state image is generated from the image. The factoid and the current state image are combined into a combined factoid, and the combined factoid is stored for sending to a media outlet for presentation on a media format.

Abstract: Devices and methods for modeling streaming data are disclosed. A method includes: receiving, by a computing device, a local graph model; determining, by the computing device, a subgraph in the local graph model; acquiring, by the computing device, an external graph model; determining, by the computing device, a plurality of alternative subgraphs in the external graph model; determining, by the computing device, a score for each of the plurality of alternative subgraphs; selecting, by the computing device, an alternative subgraph having a highest score among the plurality of alternative subgraphs; and ensembling, by the computing device, the local graph model and the alternative subgraph having the highest score.

Abstract: According to a first aspect of the present invention, there is provided a computer implemented method, a computer system and a computer program product, including training a set of exploitation models, training a set of exploration models, generating a combined exploitation and exploration heat map, and inputting the combined exploitation and exploration heat map into a convoluted neural network.

Abstract: Techniques for tailoring sampling rates for data from interactive digital properties on a feature-by-feature basis and collecting the data using the tailored sampling rates. Each feature may have an independent sampling rate irrespective of sampling rates assigned to other features. The independent sampling rates are determined based on at least one factor of predictive feature usage information based on historical feature usage information, predetermined rules, and current usage velocity of the feature. In some embodiments the independent sampling rate is influenced by the usage of an allocated resource provided to the digital property relative to a total allocation of that resource for a given time period. In some embodiments, the allocated resource is server calls to a digital data analytics server for the purposes of providing feature usage information from the interactive digital property for the performance of digital data analytics.

Abstract: In an embodiment, a method includes detecting a motion pattern in proximity to a first midair interface (MAI) device, the motion pattern being of a body part of a user. In an embodiment, the method includes converting the detected motion pattern to a simulated surface of an object projected from a shared MAI device, wherein the first MAI device and the shared MAI device each correspond to a different user. In an embodiment, the method includes causing a behavior change in the simulated surface being projected from the shared MAI device. An embodiment includes a computer usable program product. The computer usable program product includes a computer-readable storage device, and program instructions stored on the storage device.

Abstract: A hierarchy of agents is constructed from a set of agents. Each agent in the hierarchy is trained to answer a question according to a corresponding corpus associated with the agent, which contains a portion of knowledge about a subject-matter. The question is submitted to a first subset of agents, the agents in the first subset occupying a first level in the hierarchy. From a first agent in the first subset, a first answer is propagated to a second agent in a second subset of agents, the first agent computing the first answer using a first portion of knowledge about the subject-matter. to form a first morphed answer, a second answer is added to the first answer, the second answer being computed by the second agent using a second portion of knowledge about the subject-matter. The morphed answer is produced in response to the question.

Abstract: Systems and methods for customizing digital content based on consumer context data are disclosed. In embodiments, a computer-implemented method, comprises: determining, by a computing device, a goodness of fit of one or more digital content segments for a digital product based on asset success metrics indicating a trend in consumer consumption of content; identifying, by the computing device, a select digital content segment from the one or more digital content segments based on the goodness of fit; and incorporating, by the computing device, the select digital content segment into the digital product.

Abstract: Distributing computation workload among computing nodes of differing computing paradigms is provided. Compute gravity of each computing node in a cloud computing paradigm and each computing node in a client network computing paradigm within an Internet of Systems is calculated. Each component part of an algorithm is distributed to an appropriate computing node of the cloud computing paradigm and client network computing paradigm based on calculated compute gravity of each respective computing node within the Internet of Systems. Computation workload of each component part of the algorithm is distributed to a respective computing node of the cloud computing paradigm and the client network computing paradigm having a corresponding component part of the algorithm for processing.

Abstract: In an approach for acoustic modeling with a language model, a computer isolates an audio stream. The computer identifies one or more language models based at least in part on the isolated audio stream. The computer selects a language model from the identified one or more language models. The computer creates a text based on the selected language model and the isolated audio stream. The computer creates an acoustic model based on the created text. The computer generates a confidence level associated with the created acoustic model. The computer selects a highest ranked language model based at least in part on the generated confidence level.

Abstract: An adjustment specification specifies the adjustment to be made to a multimedia content, including a time characteristic. The multimedia content is matched within a tolerance to a set of comparable multimedia contents having a characteristic corresponding to the adjustment specification. An embedding model is configured and trained. Using the trained embedding model and the set of comparable multimedia contents, a set of styles is generated. Using a frame adjustment model and a style in the set of styles, a video frame of the multimedia content is adjusted. Using an audio adjustment model and the trained embedding model, an audio portion of the multimedia content is adjusted. The video frame of the multimedia content and the audio portion of the multimedia content are synchronized, producing an adjusted multimedia content adjusted according to the adjustment characteristic.

Abstract: Computer-implemented method includes developing, via a processor, a words model from a plurality of natural language text based articles relating to a subject and generating, via the processor, a static vector based upon the words model. The computer-implemented method further includes developing, via the processor, an actual articles model from actual articles, generating, via the processor, a bootstrapped vector using the actual articles model, generating, via the processor, a n-dimensional depth item using the static vector and the bootstrapped vector, and determining, via the processor, evidence based on the n-dimensional depth item. The computer-implemented method still further includes presenting, via the processor and a display, the evidence base upon an input query from a user.