Abstract: Disclosed herein are various approaches for sharing knowledge within and between organizations while protecting sensitive data. A machine learning model may be trained using training prompts querying a vector store to prevent unauthorized user disclosure of data derived from the vector store. A prompt may be received and a response to the prompt may be generated using the machine learning model based at least in part on the vector store.

Abstract: The present disclosure relates to LLM orchestration with vector store generation. An embeddings model may be selected to generate an embedding for a digital artifact. Metadata for the digital artifact may also be generated and stored in a vector store in association with the embedding. A user query may be received and categorized. One of a plurality of machine learning models may be selected based on the categorization of the user query. A prompt may be generated based at least in part on the user query, and the selected machine learning model may generate a response to the user query based at least in part on the prompt.

Abstract: Methods and related systems for operating a furnace are disclosed. In an embodiment, the method includes activating a burner assembly and a first fan of the furnace to combust fuel and air and circulate combustion gases along a flow path extending through a heat exchanger of the furnace. In addition, the method includes operating a second fan of the furnace to circulate air across an external surface of the heat exchanger of the furnace and produce a conditioned airflow. Further, the method includes monitoring one or more parameters of a motor of the second fan indicative of an airflow rate of the conditioned airflow, and deactivating the burner assembly, whereby combustion of the fuel and air in the furnace ceases, in response to the one or more parameters indicating that the airflow rate is less than a minimum airflow rate.

Abstract: A method includes obtaining behavioral source code defining logic to be performed using at least one logic device and constraints identifying data movements associated with execution of the logic. The at least one logic device contains multiple components that support at least one of: internal data movements within the at least one logic device and external data movements external to the logic device as defined by the behavioral source code and the constraints. The constraints identify characteristics of at least one of: the internal data movements and the external data movements. The method also includes automatically designing one or more data movers for use within the at least one logic device, where the one or more data movers are configured to perform at least one of the internal and external data movements in accordance with the characteristics.

Abstract: A system for facial analysis includes a camera, a data storage device and a data processing system. The camera takes video of a subject's face, and the data storage device receives and stores the video. The data processing system extracts a pose of the subject's face, and a representation of the subject's facial gesture state. The pose includes the angle and position of the subject's face. The representation includes facial keypoints that are a collection of points on the subject's face. The system then concatenates each data stream to align the data streams in time, extracts a plurality of facial syllables from the aligned data streams, and compiles the facial syllables into a series of state sequences. Based on the series of state sequences, the system extracts a behavioral fingerprint for the subject that provides a summary of the subject's state over a given period of time.

Abstract: System and methods include obtaining data, over the Internet, associated with a plurality of Wi-Fi networks each Wi-Fi network having one or more access points and each Wi-Fi network being associated with a customer of one or more service providers; aggregating and filtering the data; analyzing the aggregated and filtered data for Wi-Fi metric based alarms, each Wi-Fi metric based alarm being associated with detection of one of an offline Wi-Fi network of the plurality of Wi-Fi networks, an offline node of the Wi-Fi network, instability of the Wi-Fi network, congestion in the Wi-Fi network, and interference in the Wi-Fi network; determining the Wi-Fi metric based alarms based on the analyzing; and performing one or more actions based on the determined Wi-Fi metric based alarms.

Abstract: A method for more efficiently storing genomic includes designating multiple different data storage techniques for storing genomic data generated by a genomic pipeline. The method further identifies a file, made up of multiple blocks, generated by the genomic pipeline. The method determines which data storage technique is most optimal to store each block of the file. In doing so, the method may consider the type of the file, the stage of the genomic pipeline that generated the file, the access frequency for blocks of the file, the most accessed blocks of the file, and the like. The method stores each block using the data storage technique determined to be most optimal after completion of a designated stage of the genomic pipeline, such that blocks of the file are stored using several different data storage techniques. A corresponding system and computer program product are also disclosed.

Abstract: A method and a sensor device for evaluating residual Sulphur in a cement preheater of a cement kiln, wherein the residual Sulphur is based on the values of the fuel Sulphur content, the fuel rate of consumption, the hotmeal quality and the clinker Sulphur content. A method for evaluating blockage in a cement preheater includes evaluating the residual Sulphur in the cement preheater, determining an agglomeration rate of Sulphur compounds agglomerating on an inner surface of the cement preheater based on the residual Sulphur, and evaluating a level of blockage in at least one predetermined pathway of the cement preheater using a blockage evaluation unit, wherein the level of blockage is based on the agglomeration rate.

Abstract: Disclosed is an improved approach to implement anomaly detection, where an ensemble detection mechanism is provided. An improvement is provided for the KNN algorithm where scaling is applied to permit efficient detection of multiple categories of anomalies. Further extensions are used to optimize local anomaly detection.

Abstract: Techniques for computing global feature explanations using adaptive sampling are provided. In one technique, first and second samples from an dataset are identified. A first set of feature importance values (FIVs) is generated based on the first sample and a machine-learned model. A second set of FIVs is generated based on the second sample and the model. If a result of a comparison between the first and second FIV sets does not satisfy criteria, then: (i) an aggregated set is generated based on the last two FIV sets; (ii) a new sample that is double the size of a previous sample is identified from the dataset; (iii) a current FIV set is generated based on the new sample and the model; (iv) determine whether a result of a comparison between the current and aggregated FIV sets satisfies criteria; repeating (i)-(iv) until the result of the last comparison satisfies the criteria.

Abstract: A processor may identify that an external power source has begun powering a computing device. The processor may identify computational data in a volatile memory of the computing device. The processor may determine that the external power source does not have sufficient energy capacity to provide the computing device enough power to process the computational data at a first I/O throttling rate. The processor may increase the first I/O throttling rate to a second I/O throttling rate. The second I/O throttling rate may allow the computational data to be processed by the computing device with the energy capacity of the external power source.

Abstract: A computer-implemented method according to one embodiment includes compiling log data from all nodes of a cluster, retrieving a timestamp table stored within the cluster, optimizing the timestamp table, and adjusting the log data from all the nodes of a cluster, utilizing the timestamp table.

Abstract: A computer-implemented method according to an aspect includes training a cognitive network, utilizing metadata associated with historic data threats, inputting metadata associated with a current data threat into the trained cognitive network, identifying, by the trained cognitive network, one or more stored instances of data determined to be vulnerable to the current data threat, and adjusting one or more security aspects of the one or more stored instances of data determined to be vulnerable to the current data threat.

Abstract: A computer-implemented method, according to one embodiment, includes: receiving one or more suggestions which correspond to placement of data in storage. The one or more suggestions are used to identify portions of actual data stored in actual storage which correspond to the one or more suggestions. The first tier is configured to stripe data across the two or more shared nodes. For each of the identified portions of the actual data stored in the first tier, the one or more suggestions is further used to determine whether to transfer the given identified portion of the actual data to the second tier. Moreover, in response to a determination to transfer at least one of the identified portions of the actual data to the second tier, one or more instructions are sent to transfer the at least one of the identified portions of the actual data from the first tier to the second tier.

Abstract: A data storage system has multiple tiers of data storage including an upper tier having a lower access latency and a lower tier having a higher access latency. A storage controller of the data storage system receives, via an interface, an access request of a workload for a target file system object, where the target file system object has an associated temperature. In response to the access request for the target file system object, the storage controller accesses the target file system object in the data storage and conditions update of the associated temperature of the target file system object based on at least one of a set including the interface and the workload. The storage controller distributes a collection of file system objects including the target file system object among the multiple tiers based on respective heats of file system objects in the collection.

Abstract: A computer-implemented method, according to one embodiment, includes: receiving one or more suggestions which correspond to placement of data in storage, where the one or more suggestions are based on data workload characteristics. The one or more suggestions are used to identify portions of actual data stored in actual storage which correspond to the one or more suggestions. For each of the identified portions of the actual data stored in the first tier, the one or more suggestions is further used to determine whether to transfer the given identified portion of the actual data to the second tier. Moreover, in response to determining to transfer at least one of the identified portions of the actual data to the second tier, one or more instructions are sent to transfer the at least one of the identified portions of the actual data from the first tier to the second tier.

Abstract: Herein are timeseries preprocessing, model selection, and hyperparameter tuning techniques for forecasting development based on temporal statistics of a timeseries and a single feed-forward pass through a machine learning (ML) pipeline. In an embodiment, a computer hosts and operates the ML pipeline that automatically measures temporal statistic(s) of a timeseries. ML algorithm selection, cross validation, and hyperparameters tuning is based on the temporal statistics of the timeseries. The result from the ML pipeline is a rigorously trained and production ready ML model that is validated to have increased accuracy for multiple prediction horizons. Based on the temporal statistics, efficiency is achieved by asymmetry of investment of computer resources in the tuning and training of the most promising ML algorithm(s). Compared to other approaches, this ML pipeline produces a more accurate ML model for a given amount of computer resources and consumes fewer computer resources to achieve a given accuracy.

Abstract: An efficient cloning mechanism is provided for a distributed storage environment, where, for example, a private cloud computing environment and a public cloud computing environment are included in a hybrid cloud computing environment (on-premise object storage to off-premise computation resources), to improve computation workloads. The disclosed algorithm forms an efficient cloning mechanism in a hybrid storage environment where the read/write speed of data from the disk is not limited by its angular velocity.

Abstract: An approach for managing memory usage in cloud and traditional environments using usage analytics is disclosed. The approach may be implemented in a computer infrastructure including a combination of hardware and software. The approach includes determining that space is available within one or more tables which have schema definitions with string fields having a predefined length. The approach further includes creating a virtual table and mapping the available space to the virtual table for population by one or more records.

Abstract: Described embodiments provide systems and methods for invalidating a cache of a domain name system (DNS) information based on changes in internet protocol (IP) families. A mobile device having one or more network interfaces configured to communicate over a plurality of networks using a plurality of internet protocol (IP) families is configured to maintain a cache storing DNS information of one or more IP addresses of a first IP family of the plurality of IP families used by the mobile device for a connection to a first network of the plurality of networks. The device can detect a change in the connection of the mobile device from the first network using the first IP family to a second network using a second IP family different from the first IP family and flush at least the DNS information of one or more IP addresses of the first IP family from the cache to prevent use by the mobile device of an IP address that corresponds to an invalid cache entry.