Holistic Traffic Control Through Q-Learning and Enhanced Deep Learning for Distributed Co-Inference
Bibliographic record
Abstract
Services for AI tasks have garnered a lot of attention as an integral aspect of intelligent services in the new era.However, implementing such a system in a stable and distributed manner, while simultaneously coordinating the use of cloud computing and remote edge devices, is challenging due to the pressing need for energy and computing resources.The primary contribution of this study lies in the development of a distributed co-inference architecture that harnesses the collective intelligence of interconnected agents to optimize traffic flow in real-time.By combining Q-learning with enhanced deep learning, our approach enables traffic signals and routing decisions to adapt dynamically to changing traffic patterns and environmental conditions.The security, responsiveness, and dependability of intelligent systems deployed close to end-users are improved by deploying deep learning systems.Another critical aspect where latency and accuracy in models are traded off is deep learning model optimization.Finding the best offloading policy and model for deep learning services requires an end-to-end decision-making solution that takes into account computation-communication problems.This study presents a holistic network optimization approach for scheduling AI services based on artificial intelligence.By adjusting for differences in computational resources and network congestion, the suggested deep Q-learning technique maximizes the throughput of AI tasks in general.This research introduces a virtual queue for analyzing the system's Lyapunov stability and employs a multi-hop Directed Acyclic Graph (DAG) to explain Q-learning of Reinforcement learning-based co-inference network topology.To optimize the total task processing rate, the study develops an Optimized self-adaptive glow worm swarm optimization method (SA-GSO) based on deep Q-learning.It then proposes a prioritybased data forwarding approach for efficiency.The study concludes by simulating the distributed co-inference system's platform.We attest to the superiority of our idea by comparing it to other standards.
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 | 0.004 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| 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 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".