Abstract: Implementations are directed to processing a content object model through a ML model to provide a set of base content feature representations, processing a style object model through the ML model to provide a set of base style feature representations, executing iterations including: generating, by the ML model, a set of stylized feature representations for an initial stylized object model, the initial stylized object model having one or more adjusted parameters relative to a previous iteration, determining a total loss based on the set of stylized feature representations, the set of base content feature representations, and the sets of base style feature representations, and determining that the total loss is non-optimized, and in response, initiating a next iteration, executing an iteration including determining that the total loss is optimized, and in response providing the initial stylized object model as output of the iterative process.

Abstract: An artificial intelligence (AI) technique to process and query data pertaining to an enterprise. A user raises a request which is processed to predict a knowledge context area based on a predetermined structure of the enterprise. The knowledge context area is predicted from multiple knowledge context areas, on the basis of the received user request and a conversation history of the user in past. Further, a knowledge database is selected from multiple knowledge databases based on the user request and the predicted knowledge context. The knowledge databases include preprocessed data from multiple data sources. The knowledge database is queried on the basis of the user request related to the knowledge context to obtain a result and the result is then displayed as an output.

Abstract: A process circularity optimization system improves the circularity of industrial processes via a non-linear analysis of process attributes. A process input provides one or more target attributes to be optimized along with an objective function for improving the circularity of a manufacturing process. The input variables to be set to optimize the target attributes along with any constraints are accessed. Multiple input variable value combinations are generated via non-linear processing of the input variables and a combination of input variable values with the highest probability of success is implemented in the manufacturing process to optimize circularity.

Abstract: Systems and methods for providing an optimized customer fallout framework are disclosed. A system receives an input from a user via a digital platform, processes the input and historical data to extract a set of quantifiable features, determines an engagement stage from a plurality of engagement stages for the user based on the set of quantifiable features and an n-helix multi-dimensional model, and determines, via a deep learning model corresponding to a combination of the engagement stage and an advanced engagement stage, a set of positive drivers for the user to move from the engagement stage to the advanced engagement stage. Further, the system generate a multi-nodal network comprising a plurality of grids and aggregates the plurality of grids to generate a global grid for the determined set of positive drivers. Based on the global grid, the system dynamically generates personalized recommendations for the user.

Abstract: This document describes self-cleaning devices. In one aspect, a self-cleaning device includes a fabric having a surface covered with a photocatalyst, one or more light sources embedded in the fabric, and a triggering mechanism that activates a cleaning cycle by activating the one or more light sources. The triggering mechanism can include a pressure sensor. The triggering mechanism can be configured to activate the cleaning cycle in response to detecting a decrease in pressure being applied to the pressure sensor.

Abstract: A system may receive, from one or more data sources, one or more de-identified data sets that include de-identified personal data. The system may receive a request for a feature set of the one or more de-identified data sets, wherein the feature set includes a set of quasi-identifiers included in the de-identified personal data. The system may calculate a re-identification risk score for the set of quasi-identifiers. The system may selectively output, based on the re-identification risk score, one of: actual data, from the one or more de-identified data sets, of the feature set if the re-identification risk score satisfies a condition, or synthetic data, generated by the device from the one or more de-identified data sets, for the feature set, or a combination of the synthetic data and the actual data for the feature set, if the re-identification risk score does not satisfy the condition.

Abstract: A system includes circuitry for rewriting blockchains in a non-tamper-evident or tamper-evident operation using a key secret held in portions by multiple individually untrusted parties. The blockchains may include a series of blocks secured by integrity codes that may prevent non-tamper-evident rewrites by non-trusted parties that are not in possession of the key secret or individually-untrusted parties in possession of only a portion of the key secret. In some cases, multiple individually-untrusted parties may combine their portions into the key secret. As a group, the multiple individually-untrusted parties may perform non-tamper-evident operation with respect to at least one integrity code within the blockchain.

Abstract: A remote location monitoring system identifies a place of interest and a subset of artificial satellites that can capture images of the place of interest within a threshold period. The next available artificial satellite with image sensors that can capture images of the place of interest earliest is selected and the cache heating signal is transmitted to be stored in a cache associated with the next available image sensor. The cache heating signal activates particular image sensors for image capture and enables the transmission of meaningful images that enable monitoring the place of interest.

Abstract: The disclosure below is directed to optical/photonic devices and systems configured to control polarization state of light for selective privacy protection in optical imaging. In particular, the various embodiments disclosed below provide an application of polarization manipulation techniques with multiple optical polarization filters to enable partial or full privacy while maintaining functioning of image-based spatial mapping and localization algorithms used in video devices such as virtual reality (VR), augmented reality (AR), and robot/drone cameras. In the disclosed example approaches, the various polarization filters are configured to allow no or negligibly small amount of light from the privacy protected regions of a scene into the cameras in a privacy mode and to reject 100% of all light and enable full privacy in a block mode. The disclosed approaches also dynamically account for orientation changes during operation of the camera devices such that privacy is dynamically protected in real-time.

Abstract: In some examples, digital twin-based floor layout generation may include receiving, for a floor plan that is to be generated, an activity map that includes movement of at least one user within a digital twin of a specified area. Based on the activity map, embedding vectors may be generated for each room type of a plurality of room types in the specified area. An input boundary feature map may be received. The floor plan may be generated based on an analysis of the embedding vectors for each room type of the plurality of room types and based on an analysis of the input boundary feature map.

Abstract: A method and system for assisting program code development are disclosed. The method may include obtaining multimodal data of a meeting discussing a program code development, extracting a plurality of topics and a plurality of concepts from the multimodal data, identifying a plurality of meeting segments for the plurality of concepts. The method may further include determining a coding intent from program codes for the program code development, aligning the coding intent to a set of topics, identifying the set of concepts associated with the topic aligned with the coding intent, identifying a set of meeting segments associated with the concept. The method may further include determining an alignment metric of the meeting segment based on an alignment metric between the coding intent and the topic, and outputting one or more meeting segments for the coding intent based on alignment metrics of the one or more meeting segments.

Abstract: Methods, systems, and computer-readable storage media for receiving, by a self-healing platform within the container orchestration system, fault data that is representative of two or more error events occurring within a cluster provisioned within the container orchestration system, determining, by the self-healing platform, a set of actions to be executed in response to the two or more error events, providing, by the self-healing platform, a priority value for each error event of the two or more error events, and transmitting, by the self-healing platform, instructions to execute actions in the set of actions based on respective priority values of the two or more error events.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for recognizing speech using a spiking neural network acoustic model implemented on a neuromorphic processor are described. In one aspect, a method includes receiving, a trained acoustic model implemented as a spiking neural network (SNN) on a neuromorphic processor of a client device, a set of feature coefficients that represent acoustic energy of input audio received from a microphone communicably coupled to the client device. The acoustic model is trained to predict speech sounds based on input feature coefficients. The acoustic model generates output data indicating predicted speech sounds corresponding to the set of feature coefficients that represent the input audio received from the microphone. The neuromorphic processor updates one or more parameters of the acoustic model using one or more learning rules and the predicted speech sounds of the output data.

Abstract: In some examples, source code differential pruning-based dataset creation may include receiving source code that includes at least one vulnerability and at least one remediation that remediates the at least one vulnerability, extracting at least one remediated section, and identifying each sentence of the remediated section. A plurality of clusters may be generated based on an analysis of each identified sentence of the remediated section to determine a score with respect to a specified cluster that includes the identified sentence. Further, a determination may be made as to whether the score is greater than a specified threshold. Each identified sentence for which the score is greater than the specified threshold may be designated as a relevant sentence. An auxiliary dataset may be generated based on a plurality of relevant sentences and include at least one relevant vulnerability and at least one relevant remediation that remediates the relevant vulnerability.

Abstract: Aspects of the present disclosure provide systems, methods, and computer-readable storage media that support automated failure mode detection for rotating machinery based on vibration analysis. To illustrate, a computing device may receive vibration data from one or more sensors configured to measure vibrations of a rotating machine, such as an engine. The computing device may generate a frequency spectrum based on the vibration data (or receive the frequency spectrum) and compare the frequency spectrum to one or more predetermined frequency models to determine one or more similarity metrics. The one or more predetermined frequency models may each be associated with a respective failure mode of the rotating machine. The computing device may identify a failure mode associated with a predetermined frequency model that corresponds to a similarity metric that satisfies a threshold, and the computing device may output an indication of the identified failure mode.

Abstract: Methods, systems, and apparatus are provided for generating an image. A personalized text prompt is generated by processing an input embedding using a transformer model followed by a first fully connected neural network. The input embedding comprises a multi-dimensional embedding vector associated with a user profile and a plurality of user items. A scored label set is generated identifying a user's preferences by processing a set of attributes for the plurality of user items using a second fully connected neural network. The image is generated by processing the personalized text prompt and the scored label set using a diffusion model.

Abstract: A method and system for generating recommended data insights are disclosed. The method may include obtaining available Key Performance Indicators (KPIs) associated with an operation process from a KPI repository. The available KPIs may represent metrics that can be calculated based on data for the operation process stored in a data repository. The method may include identifying in-use KPIs comprising a subset of the available KPIs and calculating a data enthalpy metric for the operation process based on a number of the available KPIs and a number of the in-use KPIs. The method may further include obtaining an insight recommendation model trained to predict a significance of an available KPI not in use, executing the insight recommendation model to generate a KPI recommendation for the operation process based on the data enthalpy metric, and outputting the KPI recommendation via the user interface.

Abstract: A device may receive a data input that is associated with an event. The device may parse the data input to identify an input value that is associated with the event. The device may determine a probability that the input value corresponds to a feature of the event based on a configuration of the input value. The device may classify the input value as being associated with an element of the event based on the probability. The device may determine a rule profile of the input value based on the feature and the element. The device may determine a profile score associated with the data input based on the rule profile. The device may ingest, based on the profile score, the data input into a data structure. The device may determine a validation score based on a random factorization analysis of the rule profile and the input value.

Abstract: This disclosure is directed to a computer system and method to assist in identifying data feature intersection or overlap between private datasets without revealing any specific data items or data features in the datasets. Various technical components including natural language processing, lexical optimization, and encryption and key management technologies such as homomorphic encryption and secret sharing and coding, are integrated into the disclosed system and method to achieve the data feature intersection identification. Such a system and method may be employed in circumstances where data feature intersection is important for collaborative efforts between entities.

Abstract: Embodiments of this disclosure disclose a method and system for optimizing resources for cloud-based scalable distributed search data analytics service. The method may include generating a computing capacity rightsizing recommendation on the computing capacity based on the computing utilization metrics and generating a memory capacity rightsizing recommendation on the memory capacity based on the memory utilization metrics. The method may further include determining a recommended instance type for the service resource unit based on the computing capacity rightsizing recommendation and the memory capacity rightsizing recommendation. The method may further include performing a storage volume check on the service resource unit to obtain a storage volume check result.