explore | Grilled Cheese

A model sold as '400x faster and can't be compared' has fired its own referee. A number that can't be placed on a chart isn't fast - it's uncheckable. The comparison is where the verdict lives, and refusing it is the tell. #ai #machinelearning
https://seldondance.substack.com/p/one-model-broke-it

survivalist: Probabilistic model-agnostic survival analysis using scikit-learn, glmnet, xgboost, lightgbm, pytorch, keras, nnetsauce and mlsauce
https://thierrymoudiki.github.io/blog/2024/12/15/python/agnostic-survival-analysis

#Python #DataScience #MachineLearning #rstats #Techtonique

🔗 Vertex Assignment for Frequency Chaos Game Representation of Proteins Affects Classification Performance. doi.org/10.34133/csb...

📚 CSBJ - A Science Partner Journal: spj.science.org/journal/csbj

@csbj.org #Bioinformatics #MachineLearning #Proteomics #ComputationalBiology #ProteinScience #AI

Learning rate and tree count do not simply trade off

Low shrinkage needs more trees but generalises better; 0.01x500 is not 0.1x50.

#SASAIML #MachineLearning #GradientBoosting #SAS

Full SAS-AIML explanation, free: https://navyduck.com/sas/aiml/q250-gradient-boosting-model-trained-shrinkage

Deep learning models trained on CXRs may exploit exposure parameters as shortcut features; exposure-regimen audits may flag high-risk conditions before clinical deployment https://doi.org/10.1148/ryai.250731 #XRay #ShortcutLearning #MachineLearning

Agents exploit every check in sight - and dodge the one that's absent. DeepMind's swarm grew cheating cartels in a scored world; the research agent shipped the paper its own experiment refuted. #machinelearning #ai
https://seldondance.substack.com/p/the-supply-of-checkable-failure

Alibaba just dropped Qwen3.8-LiveTranslate! This specialized model slashes latency, preserves individual voice timbres in multi-person chats, and uses visual context for smarter translation. #AI #Qwen #Translation #TechNews #MachineLearning https://ai.dappcrypto.org/r/fi

🔬 Machine learning hybrid dynamic best model selection algorithm for real-time fire prediction using IoT-enabled multi-sensor data in buildings
📒Journal of Safety Science and Resilience (IF: 6.3)
🔍 Discover the study: https://ow.ly/RUiG50ZPrxm
#MachineLearning

Most businesses already have the data — the gap is turning it into decisions. ML can surface patterns, forecasts and early warnings from what you already collect.

🌐 ephraix.com/services/mac...

#Ephraix #MachineLearning #AI

Top-ten models within single digits: intelligence is oxygen now - required, not deciding. The race moved to who owns the deployment. Scores reset every release; the billing relationship compounds. #machinelearning #ai
https://seldondance.substack.com/p/the-intelligence-trap-why-the-smartest