Shallow or deep? Detecting anomalous flows in the Canadian Automated Clearing and Settlement System using an autoencoder
Bibliographic record
Abstract
Financial market infrastructures and their participants play a crucial role in the economy. Financial or operational challenges faced by one participant can have contagion effects and pose risks to the broader financial system. Our paper applies (deep) neural networks (autoencoder) to detect anomalous flows from payments data in the Canadian Automated Clearing and Settlement System (ACSS) similar to Triepels et al. (2018). We evaluate several neural network architecture setups based on the size and number of hidden layers, as well as differing activation functions dependent on how the input data was normalized. As the Canadian financial system has not faced bank runs in recent memory, we train the models on Å“normal data and evaluate out-of-sample using test data based on historical anomalies as well as simulated bank runs. Our out-of-sample simulations demonstrate the autoencoders performance in different scenarios, and results suggest that the autoencoder detects anomalous payment flows reasonably well. Our work highlights the challenges and trade-offs in employing a workhorse deep-learning model in an operational context and raises policy questions around how such outlier signals can be used by the system operator in complying with the prominent payment systems guidelines and by financial stability experts in assessing the impact on the financial system of a financial institution that shows extreme behaviour.Â
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 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.001 | 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".