Datasets:
File size: 3,572 Bytes
6c1c833 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 | {
"field_id": "F018",
"field": "anomaly detection",
"directions": [
{
"direction_id": "G0",
"name": "Normal-content restoration for anomaly detection",
"scope": "Obtain a normal counterpart of an observation through reconstruction, completion, or denoising, and detect or localize anomalies from the differences."
},
{
"direction_id": "G1",
"name": "Temporal expectation and pattern deviations",
"scope": "Use expected states of temporal behavior or observed recurring patterns as references to identify anomalous periods, trajectories, or subsequent observations."
},
{
"direction_id": "G2",
"name": "Distributional and sample-geometric outlier scoring",
"scope": "Directly assign anomaly scores based on probabilistic evidence for observations under a normal distribution, or on their sparse or isolated positions relative to other samples."
},
{
"direction_id": "G3",
"name": "Anomaly-discriminative feature learning and references",
"scope": "Make a feature space distinguish normal structure from deviations, and use normal feature references or learned representations for detection and localization."
},
{
"direction_id": "G4",
"name": "Learning with limited anomaly evidence",
"scope": "When real anomaly examples are absent or scarce, construct surrogate anomalies, use a few known anomalies, or selectively obtain expert feedback to learn more targeted detectors."
},
{
"direction_id": "G5",
"name": "Structural and relational anomaly detection",
"scope": "Use relationships among entities, attributes, or concurrently varying variables to determine whether structural or relational behavior deviates from normal, and localize the deviation where applicable."
},
{
"direction_id": "G6",
"name": "Video behavior anomaly recognition and localization",
"scope": "Recognize anomalous behavior from actions, appearance, scenes, and their spatiotemporal relationships in video, and identify anomalous segments when needed."
},
{
"direction_id": "G7",
"name": "Anomaly capability transfer across target settings",
"scope": "When target-specific anomaly training is limited or source and target conditions differ, transfer existing knowledge to maintain anomaly recognition and localization capability in the target setting."
},
{
"direction_id": "G8",
"name": "Sustained anomaly monitoring under deployment constraints",
"scope": "Maintain timely and usable anomaly monitoring on continuously arriving data despite changing conditions or computational, storage, or device constraints."
},
{
"direction_id": "G9",
"name": "Anomaly explanation and attribution",
"scope": "Provide usable feature attributions, explanations of causes, or counterfactual explanations of what would make flagged anomalies normal."
},
{
"direction_id": "G10",
"name": "Detector configuration and combination",
"scope": "When candidate detectors or configurations are uncertain, select, tune, or combine candidate approaches to obtain more reliable anomaly judgments."
},
{
"direction_id": "G11",
"name": "Anomaly detection evaluation and assessment",
"scope": "Use benchmarks, metrics, or systematic comparisons to assess anomaly detectors’ capabilities, limitations, and evaluation conclusions, rather than directly selecting a detector for a particular dataset."
}
]
}
|