接続

ACM CoNEXT Experiment Design

brycewang-stanfordbybrycewang-stanford
•
1.1k
•
学習とドキュメンテーション

Audits and designs rigorous experimental evaluations for ACM CoNEXT networking papers, ensuring testbed fidelity, tuned baselines, and statistical rigor.

ACM CoNEXT Experiment Design guides researchers in crafting systems-and-measurement evaluations that meet the rigorous standards of the ACM CoNEXT networking conference. It helps map empirical claims to appropriate real-world target platforms—such as hardware testbeds, deployments, and trace-driven replays—while enforcing honest baseline tuning, statistical uncertainty reporting (CDFs, tail metrics, confidence intervals), trace provenance, and contamination-aware ML-for-networking ablations to pre-empt major revision risks.

主な機能

01Claim-to-evidence matching across real networking testbeds, switches, and deployments
02Auditing baseline configurations to prevent strawman comparisons
03Trace, vantage point, and configuration provenance tracking
041,074 GitHub stars
05Statistical uncertainty verification including CDF distributions and confidence intervals
06Contamination-aware ML-for-networking ablations and model-swap validation

ユースケース

01Designing realistic hardware testbed evaluations for novel networking protocols or switch mechanisms
02Auditing experimental setups prior to submitting to ACM CoNEXT to avoid revision traps
03Structuring ML-for-networking evaluation pipelines with proper data splits, ablations, and heuristic baselines