A Novel Flight Dynamics Modeling Using Robust Support Vector Regression against Adversarial Attacks
Bibliographic record
Abstract
An accurate Unmanned Aerial System (UAS) Flight Dynamics Model (FDM) allows us to design its efficient controller in early development phases and to increase safety while reducing costs. Flight tests are normally conducted for a pre-established number of flight conditions, and then mathematical methods are used to obtain the FDM for the entire flight envelope. For our UAS-S4 Ehecatl, 216 local FDMs corresponding to different flight conditions were utilized to create its Local Linear Scheduled Flight Dynamics Model (LLS-FDM). The initial flight envelope data containing 216 local FDMs was further augmented using interpolation and extrapolation methodologies, thus increasing the number of trimmed local FDMs of up to 3,642. Relying on this augmented dataset, the Support Vector Machine (SVM) methodology was used as a benchmarking regression algorithm due to its excellent performance when training samples could not be separated linearly. The trained Support Vector Regression (SVR) predicted the FDM for the entire flight envelope. Although the SVR-FDM showed excellent performance, it remained vulnerable to adversarial attacks. Hence, we modified it using an adversarial retraining defense algorithm by transforming it into a Robust SVR-FDM. For validation studies, the quality of predicted UAS-S4 FDM was evaluated based on the Root Locus diagram. The closeness of predicted eigenvalues to the original eigenvalues confirmed the high accuracy of the UAS-S4 SVR-FDM. The SVR prediction accuracy was evaluated at 216 flight conditions, for different numbers of neighbors, and a variety of kernel functions were also considered. In addition, the regression performance was analyzed based on the step response of state variables in the closed-loop control architecture. The SVR-FDM provided the shortest rise time and settling time, but it failed when adversarial attacks were imposed on the SVR. The Robust-SVR-FDM step response properties showed that it could provide more accurate results than the LLS-FDM approach while protecting the controller from adversarial attacks.
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 | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
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".