Skill UI

A comprehensive guide for designing and auditing empirical experiments in signal processing, audio, and speech domains. It emphasizes matching standard metrics (e.g., WER, SI-SDR, PSNR) to specific tasks, establishing current strong baselines, and systematically sweeping operating conditions (like SNR or reverberation time) to ensure robust, reliable, and publication-ready results.

A methodology guide for structuring high-stakes academic research experiments, particularly for top-tier CV conferences (like ICCV). It emphasizes systematic planning under strict deadlines, covering benchmark drift audits, ensuring fairness by accounting for pretraining confounds, structuring ablations to isolate mechanisms, and strategically sequencing runs to maximize impact before the submission deadline.

This guide provides comprehensive standards for designing, executing, and reporting systems experiments for top-tier academic venues (like ICDE). It emphasizes methodological rigor, including mandatory ablation studies, reporting variance and tail latency (p99), demonstrating scaling trends (curves over points), disclosing cost trade-offs, and ensuring full hardware and software reproducibility to withstand adversarial review.

A comprehensive guide for authors preparing submissions for ICDE conferences. It details how to properly segment content into the main 12-page body versus the supplementary material, ensuring the core argument remains self-contained while providing reviewers with runnable artifacts, extended experiments, and full technical documentation.

A comprehensive guide for researchers designing and auditing empirical evaluations for data mining papers (e.g., ICDM). It teaches how to structure a trustworthy evaluation by defining operational tasks, conducting ablation studies to isolate mechanisms, proving scalability, and implementing known-truth checks to ensure findings are valid and not mere artifacts.

This guide provides rigorous guidelines for structuring theoretical papers submitted to the International Conference on Database Theory (ICDT). It emphasizes that proofs and formal evidence are the primary contribution, not benchmarks or experiments. It teaches authors how to match their specific theoretical claims (e.g., complexity, expressiveness) to the appropriate academic evidence, ensuring that any proposed experiment is proportional, honest, and merely supplements the core theoretical results.

This guide provides comprehensive best practices for designing and auditing empirical experiments for top-tier ML conferences. It covers necessary ablations, robustness checks (e.g., data shift, seeds), fair baselines, and detailed compute reporting to ensure that reported gains are scientifically genuine, reproducible, and address common reviewer skepticism.

This guide helps authors structure lengthy academic manuscripts, ensuring the main text maintains a clear, focused narrative while relegating supportive, detailed evidence to the appendices or supplementary materials. It outlines the 'determinative' principle, advising where to place hyperparameters, extended experiments, proof details, and code links to meet high journal standards and maximize readability for reviewers.

This guide provides a comprehensive framework for stress-testing Machine Learning experimental evidence before submission or rebuttal. It covers critical areas such as establishing strong baselines, conducting mechanism-isolating ablations, reporting variance, checking for data leakage, and disclosing compute costs. Use this to ensure your ML claims are scientifically sound, robust, and fair against rigorous academic scrutiny.

A comprehensive guide for designing and reporting rigorous experimental results in robotics research. It details the 'evidence ladder' (from simulation to real-world deployment), statistical best practices (e.g., sample size, confidence intervals), ensuring baseline fairness, and quantifying sim-to-real transfer gaps. Essential reading for researchers submitting to top-tier robotics conferences.

A comprehensive guide for designing, executing, and reporting high-stakes research studies in Software Engineering (SE), tailored to meet the stringent standards of top academic conferences like ICSE. It covers best practices for evidence-claim matching, advanced statistical rigor (non-parametric tests, effect sizes), handling AI model ablations, and implementing threats-driven design to ensure findings are robust, valid, and maximally reproducible.

A comprehensive methodology guide for designing and auditing empirical studies in software engineering, specifically focused on maintenance, evolution, and change impact. It outlines best practices for ensuring evidence is proportional to the claim, managing data provenance (SHAs, criteria), handling statistical confounds (survivorship, effect sizes), and integrating rigorous LLM evaluation protocols.