AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

A Deep Reinforcement Learning Perspective on Internet Congestion Control

Nathan Jay, Noga Rotman, Brighten Godfrey, Michael Schapira, Aviv Tamar

ICML 2019 · Reinforcement learning › Visual and representation RL · approach: Scalable RL systems and environments · 162 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Train congestion control policy with deep RL (PPO) on simulated links, with linear reward (throughput, latency, loss)Aurora · Mininet/Pantheon robustness testslink utilization and self-inflicted latencyImproved
Use discount factor gamma=0.99 instead of 0.5 or 0.0 (ablation)Aurora training · Training rewardlearning speedImproved
Increase history length from 1 to 2 MIs (ablation)Aurora · Training rewardtraining rewardImproved
Aurora trained on dynamic-like links versus TCP CUBIC on dynamic linksAurora · Dynamic links (16-32 Mbps random)throughput nearly matching BBR with significantly better latencyImproved
Replace linear-model / bandit formulations with a small 2-hidden-layer neural network (32→16, tanh) (ablation)Aurora · Training rewardtraining rewardImproved
Aurora with 1% random loss vs TCP CUBICAurora · Single flow 30 Mbps with 1% random lossthroughputImproved
Aurora on fairness with TCP or other protocolsAurora · Fairness (not evaluated)fairnessNo clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy