SVEA: A Small-scale Benchmark for Validating the Usability of Post-hoc Explainable AI Solutions in Image and Signal Recognition
Bibliographic record
Abstract
Novel solutions in the area of Explainable AI (XAI) have made a significant breakthrough in increasing the trust of end-users in Machine Learning (ML) models. However, validating the performance of these solutions remains a challenging task. In this work, we focus on evaluating the methods that attribute a model’s decision to their input features. The prior metrics on this topic fail to consider multiple properties that a usable explainability solution should satisfy. Also, conducting experiments to assess the concreteness of the explanations provided by these solutions in large-scale datasets consumes excessive time and resources. To overcome these shortcomings, we propose the Small-scale Visual Explanation Analysis (SVEA) benchmark, which comprises the recent minimal MNIST-1D dataset. Our proposed benchmarking tool aids the practitioners and researchers to perform experiments on the Explainable AI methods without the need to access expensive computational devices. Furthermore, we offer a framework to evaluate various characteristics of the state-of-the-art XAI methods and include several widely used interpretability solutions in the SVEA benchmark to perform a thorough analysis of their completeness and understandability. The results obtained from our proposed evaluation metric suggest that specific approaches lack the ability to transfer the chosen model’s understanding to a second interpretable model by the explanations generated. The users can replicate our experiments within few minutes before working extensively on other larger datasets, thereby saving a lot of experimental time and effort.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| 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 teacher head, 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".