Abstract:
A method for early warning is provided. The method clusters (810) normal historical data of normal cars into groups based on the car subsystem to which they belong. The method extracts (820) (i) features based on group membership and (ii) feature correlations based on correlation graphs formed from the groups. The method trains (830) an Auto-Encoder and Auto Decoder (AE&AD) model based on the features and the feature correlations to reconstruct the normal historical data with minimum reconstruction errors. The method reconstructs (840), using the trained AE&AD model, historical data of specific car fault types with reconstruction errors, normalizes the reconstruction errors, and selects features of the car faults with a top k large errors as fault signatures. The method reconstructs (850) streaming data of monitored cars using the trained AE&AD model to determine streaming reconstruction errors, comparing the streaming reconstruction errors with the fault signatures to predict and provide alerts for impending known faults.
Abstract:
Systems and methods for data fusion and analysis of vehicle sensor data, including receiving a multiple modality input data stream from a plurality of different types of vehicle sensors, determining latent features by extracting modality-specific features from the input data stream, and aligning a distribution of the latent features of different modalities by feature-level data fusion. Classification probabilities can be determined for the latent features using a fused modality scene classifier. A tree-organized neural network can be trained to determine path probabilities and issue driving pattern judgments, with the tree-organized neural network including a soft tree model and a hard decision leaf. One or more driving pattern judgments can be issued based on a probability of possible driving patterns derived from the modality-specific features.
Abstract:
Systems and methods for defect detection for vehicle operations, including collecting a multiple modality input data stream from a plurality of different types of vehicle sensors, extracting one or more features from the input data stream using a grid-based feature extractor, and retrieving spatial attributes of objects positioned in any of a plurality of cells of the grid-based feature extractor. One or more anomalies are detected based on residual scores generated by each of cross attention-based anomaly detection and time-series-based anomaly detection. One or more defects are identified based on a generated overall defect score determined by integrating the residual scores for the cross attention-based anomaly detection and the time-series based anomaly detection being above a predetermined defect score threshold. Operation of the vehicle is controlled based on the one or more defects identified.
Abstract:
The present invention provides an information processing device that improves the detectability of system errors. This information processing device includes: a means that generates a state graph based on relationship change information indicating a change in the relationship between a plurality of elements included in a system, the state graph having the elements as the vertices thereof and the relationship between the elements as the sides thereof; a means that generates a normal model having the state graph as a set of conditions to be fulfilled during normal system operation, based on the relationship change information; and a means that detects system errors and outputs error information indicating detected errors, based on the state graph and the normal model.
Abstract:
The present invention provides an information processing device that outputs information including the data transmission relationship between elements constituting an information processing system, the information indicating the state of the information processing system. The information processing device includes a graphing means for generating a relationship graph based on an event log indicating the behavior of each of a plurality of processes operating in the system, the relationship graph having the processes as the vertices thereof and having the data transmission relationship between the vertices as the sides thereof; and a graph output means for outputting the generated relationship graph.
Abstract:
Provided are a rule distribution apparatus, an event processing system, a rule distribution method, and a rule distribution program, with which it is possible to prevent bottlenecking of specific resources in a network, thereby maintaining high service quality. The distribution apparatus is equipped with: a rule receiving unit 11, which receives processing rules that specify an action requested by an application and specify an event condition for execution of the action; and an event generation unit 12, which generates sub-rules that determine all or a portion of one or more attribute conditions included in the event condition of the processing rule, or that change the timing for sending an event, and which generates configuration rules that include an event condition specifying the timing for registration of a sub-rule in a terminal sending an event, and which sends the configuration rules and the sub-rules to a processing apparatus that is capable of communicating with the terminal and is capable of executing the event processing.
Abstract:
This invention provides an information processing device that can more appropriately extract anomaly influence processes even if anomalies are found at multiple points. Said information processing device includes a reach-extent extraction means and a shared-extent extraction means. Using both a relationship graph representing the relationships between a plurality of elements in a system and location information that indicates, on said relationship graph, a plurality of locations in the system where anomalies have been detected, the reach-extent extraction means starts at each of said locations and extracts a reach extent consisting of a path in the relationship graph including the set of elements that are directly or indirectly related to the location in question. The shared-extent extraction means extracts an anomaly influence process by extracting an extent that is shared among at least a prescribed number of the plurality of paths in the relationship graph that were extracted as reach extents.