A Machine Learning-Based Toolbox for P4 Programmable Data-Planes
Bibliographic record
Abstract
Intelligent data-planes (IDPs) can enhance network service performance and adaptation speed by executing one or more machine learning (ML) models directly on the served flows. The real-time ML inference enables line-speed decision-making for some traffic management functionalities. Due to the inherent scarcity of both the computational and memory resources and the strict high-speed per-packet processing demands, existing IDP deployments either realize only a limited set of ML models such as decision trees, or require substantial modifications in the switch hardware. In this paper, we propose INQ-MLT, a novel ML-based management toolbox to address the aforementioned limitations. INQ-MLT delegates the task of training various ML models to the control-plane. The latter adopts a tailored quantization-aware training process to compensate for the effect of precision loss resulting from quantization. The toolbox then employs a quantization mechanism to transform the trained ML model parameters (e.g., weights and activations) from floating-point representations to compact low-precision fixed integer values that can be easily processed and stored in the data-plane. Finally, the trained model is deployed into the IDP pipeline by restricting all its inference operations to basic arithmetic operations. To analyze the performance of INQ-MLT, we quantify the accuracy loss resulting from the quantization step through rigorous theoretical analysis. A proof-of-concept implementation of the proposed toolbox is developed using P4-based software switches. Experiments on two use-cases demonstrate that the deployed quantized models have almost no loss of accuracy when compared to their floating-point counterparts.
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 | 0.001 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.003 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.007 | 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".