CONTROLLING MODEL TRUST WITH COMPACTLY SUPPORTED SMOOTH RBF
Bibliographic record
Abstract
Building models for any kind of complex process is an important tool of today's applied computer science. There are many situations where the trust in the model varies over the input space, and where the amount of trust or confidence should significantly affect the behaviour of the model and the resulting decisions (this applies when the model is used within some decision process, e.g., in a control or optimization task). In this paper, we will focus on special one-sided situations where overestimating the true process is considered critical, while underestimating is tolerable (or conversely). We introduce a new type of radial basis function, the confidence term, with the following properties: (a) it is smooth, i.e., infinitely differentiable and (b) compactly supported. We show how one-sided trust control can be achieved for any kind of model by a simple multiplication with the confidence term. To demonstrate the power and flexibility of our approach, two quite different applications are presented, both of which are practically relevant. One is model-based optimization with constraints, where we have to be careful not to narrow the search space too quickly, until we can trust the constraint model. This requires imposing a low confidence on the constraint model until enough data is available. In the other application, active learning with multiple point queries, we need to achieve the opposite and impose a high value of trust in regions that have been already explored.
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 | 0.005 | 0.021 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.002 |
| Scholarly communication | 0.002 | 0.003 |
| Open science | 0.002 | 0.003 |
| Research integrity | 0.002 | 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".