Real-time Outlier Detection Over Streaming Data
Bibliographic record
Abstract
Designing outlier detection algorithms over streaming data involves several issues such as concept drift, temporal context, transience, uncertainty, etc. Moreover, to produce results in real-time with limited memory resources, the processing of such data must occur in an online fashion. Therefore, real time detection of outliers on streaming data faces more challenges than performing the same task on batches of data. Several methods have been proposed to detect outliers over streaming data, among which a sliding window technique is frequently used. In this technique, only a chunk of data is kept in memory at each point in time and used to build predictive models. The size of the data in memory simultaneously is referred to as the size of a sliding window. The correctness of the outlier detection results depends largely on the choice of window size. Other similar techniques exist but most of them fail to address the properties of streaming data, and thus produce results exhibiting poor accuracy. In this paper, we present an online outlier detection algorithm, that addresses the aforementioned challenges. The proposed algorithm adopts the sliding window technique, however efficiently mines in memory a statistical summary of previous observed data, which contributes to the prediction of incoming data. It further addresses the concept drift problem that exists in streaming data. We evaluated the accuracy of our algorithm on both synthetic and real-world datasets. Results show that the proposed method detects outliers over streaming data with higher accuracy than SOD_GPU algorithm proposed in 7516110, even when concept drifts occur. The algorithm does not require a secondary memory for processing and is further accelerated using CUDA GPU.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.008 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.002 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".