A curated selection of real publications on AI and data science — each reviewed and attributed to its original authors. No summaries, no paraphrasing.
Papers selected for clarity, methodological rigor, and relevance to practitioners working in AI and data science today.
Kaplan et al. mapped how model performance changes with compute, data, and parameters — and found that scaling any one without the others produces diminishing returns. The paper changed how labs plan training runs.
The relationships they identified still hold across architectures that came after, which makes this one of the more durable empirical findings in the field.
Read paperLundberg and Lee formalized SHAP values, giving practitioners a consistent way to explain individual predictions from any model. It bridges game theory and machine learning in a way that actually works in production.
Read paperThe AlphaGo paper from DeepMind combined Monte Carlo tree search with deep reinforcement learning to beat professional Go players. The methodology has since been adapted far beyond games.
Read paperPapers here span the core disciplines that shape how AI systems are built, evaluated, and understood. Each area has its own active research community and its own set of open problems.
Transformer variants, attention mechanisms, and efficiency improvements across model families.
Dataset construction, curation quality, and how training data composition affects model behavior.
Benchmark design, evaluation methodology, and measuring what models actually learn versus memorize.
Formal approaches to specifying objectives, reward hacking, and interpretability as a safety tool.
Generalization theory, bias-variance trade-offs, and the mathematics behind why models work.
Papers bridging research and practice — forecasting, anomaly detection, and real-world deployment.
Papers reviewed here come from researchers across academia and industry. Attribution is always complete — no anonymous summaries, no uncredited ideas.
Associate Professor, Kyiv Polytechnic Institute
NLPResearch Engineer, ML platform team
InfrastructurePhD candidate, data systems lab
EvaluationIndependent researcher, reinforcement learning
RLData scientist, applied forecasting
Time SeriesPostdoctoral researcher, interpretability
XAI