A Geometric Perspective on ML Safety Assurance
Bibliographic record
Abstract
Some people claim AI-ML suffers from a reliability glass ceiling effect, around 10e-2 per inference, that makes it incompatible with safety-criticality by several orders of magnitude. Others advocate that safety nets and development assurance will overcome this gap so that there is no real concern indeed. We propose an explanation to the reliability plateauing phenomenon based on geometry of approximant adjustment, and on ML verification practices. We advocate the need for a new field we coined as HR ML (Highly Reliable) and UHR ML (Ultra Highly Reliable). Relying on Topological Data Analysis in high dimensions, its aim is to supplement data-science pointbased verification with volume-based verification in order to meet the needed 10e-5 / inf. error rates (and beyond). We argue that process-based ML assurance and safety monitors alone will not overcome the reliability barrier. Our HR-ML concept for safety-related applications is a research proposition at the confluence of ML assurance and system assurance.
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 | 0.012 | 0.045 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.004 | 0.002 |
| Science and technology studies | 0.002 | 0.022 |
| Scholarly communication | 0.006 | 0.013 |
| Open science | 0.003 | 0.006 |
| Research integrity | 0.003 | 0.007 |
| Insufficient payload (model declined to judge) | 0.009 | 0.002 |
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".