FedVC: Virtual Clients for Federated Autonomous Driving With Imbalanced Label Distribution
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Federated learning in autonomous driving safeguards the privacy of individual vehicles during collaborative training by avoiding the exchange of raw data. These vehicles often suffer from imbalanced label distribution, making the federated learning model developed from them critical for safety. Most current solutions addressing label imbalance in federated learning are oriented towards classification tasks. However, there is limited research focused on prediction tasks involving imbalanced datasets relevant to autonomous driving. To address the identified gap, we introduce an innovative peer-to-peer federated learning framework called FedVC. This framework includes the use of virtual clients to enhance the ability of the collaborative training process to recognize the global data distribution effectively. It strategically samples relevant segments of local data from various clients for each training round and controls the execution time for backpropagation associated with each virtual client. Importantly, FedVC constructs a global perspective by utilizing metadata from each client, thereby maintaining data privacy while overcoming the challenge of label imbalance in autonomous driving prediction tasks, all without the need to share any raw datasets. Experimental results demonstrate that FedVC outperforms classical FedAvg and the most recent methods at three steering angle prediction datasets with different levels of imbalanced label distribution. The source code is publicly available at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/aikedaerC/FedVC</uri>.
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.
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 | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| 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 it