Abstract: A method for detecting novelty using an encoder and a decoder comprises: feeding a first input into the encoder and processing the first input through a plurality of encoder layers to generate a first encoded input, wherein processing the first input comprises generating a first intermediate encoded input prior to generating the first encoded input, feeding the first encoded input from the encoder into the decoder and processing the first encoded input through a plurality of decoder layers to generate a first reconstructed output, feeding the first reconstructed output from the decoder as a second or subsequent input into the encoder and processing the first reconstructed output through the plurality of encoder layers, wherein processing the first reconstructed output comprises generating a second intermediate encoded input from the one of the encoder layers, and detecting a novelty based on the first intermediate encoded input and the second intermediate encoded input.

Abstract: The present disclosure relates applying a survival analysis to model when a particular recipient will view an electronic message. For example, one or more embodiments train a survivor function to model the time that will elapse, on a continuous scale, before a recipient will open an electronic message. For example, one or more embodiments involve accessing analytics training data and extracting a first set of features affecting the time that elapsed before past recipients opened an electronic message and a second set of features affecting whether the recipients opened the electronic message at all. The system then generates a mixture model modified survivor function and determines the effect of each feature set on its corresponding outcome to learn parameters for the mixture model modified survivor function.

Abstract: Prediction systems and methods are provided. The system obtains a first social media data pertaining to a first set of users, filters the first social media data to obtain a filtered social media data, generates a word embedding matrix including co-occurrence words each represented as a vector having a context, aggregates vectors pertaining each social data to obtain a first set of vectors, and trains machine learning technique(s) (MLTs) using the first set of vectors and context of the first set of vectors. The system further obtains a second social media data pertaining to a second set of users, and performs filtering, word embedding matrix generation, and aggregation operations to obtain a second set of vectors, and further applies the trained MLTs on the second set of vectors and context associated with the second set of vectors to predict age and gender of the second set of users.

Abstract: Techniques for identifying fraudulent transactions are described. In one example method, an operation sequence and time difference information associated with a transaction are identified by a server. A probability that the transaction is a fraudulent transaction is predicted based on a result provided by a deep learning network, where the deep learning network is trained to predict fraudulent transactions based on operation sequences and time differences associated with a plurality of transaction samples, and where the deep learning network provides the result in response to input including the operation sequence and the time difference information associated with the transaction.

Abstract: A neural network embodiment comprises an input layer, an output layer and a filter layer. Each unit of the filter layer receives a filter layer input from a single preceding unit via a respective filter layer input connection. Each filter layer input connection is coupled to a different single preceding unit. The filter layer incentivizes the neural network to learn to produce a target output from the output layer for a given input to the input layer while simultaneously learning weights for each filter layer input connection. The weights learned cause the filter layer to reduce a number of filter layer units that pass respective filter layer inputs as non-zero values. When applied as an initial internal layer between an input layer and an output layer, the filter layer incentivizes the neural network to learn which neural network input features to discard to produce the target output.

Abstract: Electronic sensing systems and methods are disclosed. The electronic sensing system (ESS) receive an olfactory product and one or more smell characteristics of the olfactory product are detected and extracted by identifying a headspace of the olfactory product. A comparison of the extracted smell characteristics with one or more smell characteristics associated with a historic training data stored in a database is performed and a match between the extracted smell characteristics and the one or more smell characteristics associated with the historic training data is determined using machine learning technique(s). Further, the ESS generates a report for the olfactory product comprising at least one of type of the consumable, name of the olfactory product, a status of the olfactory product, an age of the olfactory product, and a decaying index, and classifies the olfactory product into one or more categories based on the report and/or the historic training data.

Abstract: A computer-implemented method comprising: training a pre-trained neural network that comprises: an input layer; a plurality of hidden layers, wherein each of the plurality of hidden layers has one or more nodes, wherein each of said one or more nodes has an associated weight trained based on data from a source domain; and an output layer. Training the pre-trained neural network comprises: introducing at least one additional layer to the plurality of hidden layers, wherein said additional layer has one or more nodes having associated weights; keeping weights of the nodes in the plurality of hidden layers of the pre-trained neural network unchanged; inputting data from a target domain to the input layer; and adjusting weights of the one or more nodes in the at least one additional layer based on features obtained at the output layer.

Abstract: An end-to-end cloud-based machine learning platform providing personalized game player experiences. Data lineage is generated for all transformed data for generating feature ETLs, and for training machine learning models. That data is used to understand the performance of off-line and online recommender systems for the personalization of the game player experiences. To that end, the platform pipeline provides the life cycle of the transformed data to a self-healing system that compare it to the life cycle of the user interactions. By comparing the two life cycles, the self-healing system can automatically provide a diagnostic, and it can also automatically provide an action if the performance of the model predictions has changed over time.

Abstract: An input data set for a model for estimating a missing value of a sensor in a sensor network is determined. The input data set includes one or more sensor readings selected according to a temporal-spatial parameter which is dynamic and specific to the sensor. Then the missing value of the sensor is estimated using the determined input data set as an input to the model.

Abstract: A method to present recommended operator actions of a suggestion system and to provide interaction with the suggestion system. Recommendations for two or more software applications are generated. Display and control elements for the recommendations of the two or more software applications are jointly displayed and an activation of one of the jointly displayed display and operating elements is detected. The software application assigned to the activated display and control element is opened and display and control elements are jointly displayed, each of which represent one of several recommended operator actions for the opened software.

Abstract: The technology disclosed provides a so-called “joint many-task neural network model” to solve a variety of increasingly complex natural language processing (NLP) tasks using growing depth of layers in a single end-to-end model. The model is successively trained by considering linguistic hierarchies, directly connecting word representations to all model layers, explicitly using predictions in lower tasks, and applying a so-called “successive regularization” technique to prevent catastrophic forgetting. Three examples of lower level model layers are part-of-speech (POS) tagging layer, chunking layer, and dependency parsing layer. Two examples of higher level model layers are semantic relatedness layer and textual entailment layer. The model achieves the state-of-the-art results on chunking, dependency parsing, semantic relatedness and textual entailment.

Abstract: Provided are techniques for new command assistance. A score is generated for each of a plurality of users based on a command execution history of a command by each of the plurality of users. A command template with the command and parameters of the command is generated in an order based on the command execution history of a user of the plurality of users having a high score. In response to receiving a portion of a new command, suggestions are provided for automatically completing the command using the command template.

Abstract: A system, method, and wireless earpieces for implementing a virtual assistant in response to user preferences. User preferences associated with a user of the wireless earpieces are received. Data and information about the user and an environment of the user are automatically captured by the wireless earpieces based on the user preferences. A determination is made whether to provide automatic assistance to the user based on the user preferences utilizing the virtual assistant of the wireless earpieces. The automatic assistance is generated through the virtual assistant of the wireless earpieces utilizing the data and the information. The automatic assistance is communicated to the user through the virtual assistant of the wireless earpieces.

Abstract: Display of supplemental content on a smartwatch wearable mobile device. A method establishes, on supplemental physical display plane(s) of a smartwatch wearable mobile device, virtual display plane(s). The smartwatch includes a primary physical display plane and the supplemental physical display plane(s). The supplemental physical display plane(s) at least partially surround the primary physical display plane. The method displays primary content on the primary physical display plane and supplemental content on the virtual display plane(s). The method also receives touch sensor input from a touch sensor of the smartwatch and maps the touch sensor input to interface element(s) of the supplemental content on the virtual display plane(s). The method selects the interface element(s) based on the mapping the touch sensor input to the interface element.

Abstract: An apparatus and method for moving thumbnails on a screen is provided. In a case where thumbnails desired by a user among a displayed list of thumbnails are moved by a drag-and-drop operation, if the number of thumbnails as movement targets is large, it takes time to perform a selection operation. Where one of a plurality of thumbnails displayed as a list is dragged and dropped by a user, based on a drop position, a thumbnail group including the dragged thumbnail and thumbnails positioned on a first side of the dragged thumbnail or on a second side, opposite the first side, of the dragged thumbnail among the displayed list of thumbnails is determined as a movement target for subsequent movement thereof.

Abstract: In some implementations, a computing device can present a dynamic street scene overlay when presenting a map view on a display of the computing device. The dynamic street scene overlay can be presented such that a user can clearly view both the dynamic street scene and the map view. The dynamic street scene can be dynamically adjusted in response to the user manipulating the map view to a different location. The dynamic street scene can be presented such that the objects in the images of the dynamic street scene have a three-dimensional look and feel. The dynamic street scene can be presented such that the dynamic street scene does not prevent the user from viewing and interacting with the map view.

Abstract: Example implementations are directed to systems and methods for a context aware conversational agent for self-learning. In an example implementation, a method includes generating a journaling model based on activity data and engagement data associated with one or more tasks of a user. The journaling model uses machine-learning to identify a context pattern using the activity data, and maps performance associated with the one or more tasks based at least on the engagement data. The method adaptively provides a prompt to gather additional engagement data based on the context pattern in view of real-time activity data, where the prompt is generated based on the journaling model. The journaling model is updated based on the additional engagement data.

Abstract: A system described herein may provide a mechanism for decoupling user interfaces from backend input processing systems by receiving user input that was provided to a user interface, identifying a label (e.g., a label that was defined as being associated with an interactive element via which the user input was received) associated with the user input, and identifying one or more actions to perform based on the received label.

Abstract: An artificial reality system is described that renders, presents, and controls user interface elements within an artificial reality environment, and performs actions in response to one or more detected gestures of the user. The artificial reality system includes an image capture device configured to capture image data representative of a physical environment, a head-mounted display (HMD) configured to output artificial reality content, a gesture detector configured to identify, from the image data, a gesture comprising a configuration of a hand that is substantially stationary for at least a threshold period of time and positioned such that a thumb of the hand and at least one other finger form approximately a circle or approximately a circular segment, a user interface (UI) engine to generate a UI element in response to the identified gesture, and a rendering engine to render the UI element as an overlay to the artificial reality content.

Abstract: Disclosed herein are system, method, and computer-readable storage medium embodiments for ordered macro building. An embodiment may include operations of displaying, via a user interface, a first user interface object and a second user interface object, populating a macro building pane comprising a first macro building object and a second macro building object, in response to the first user interface object receiving a first interaction and in response to the second user interface object receiving a second interaction, and compiling a macro with at least the first interaction and the second interaction, the first interaction being with a first application and the second interaction being with a second application.