Machine Learning

A practical comparison of large language models and classical machine learning, with guidance on when each approach fits a business problem.

TL;DR for operators A team has 80,000 instruction-response examples but enough resources to fine-tune on only 8,000. Selecting the examples with the highest model loss may prioritize genuinely missing knowledge, but it can also prioritize noise. Random selection and conventional quality ratings do not necessarily reveal what the current model can already retrieve. ...

TL;DR for operators An accelerator override during adaptive cruise control (ACC) operation can contain more information than a simple indication that the driver wanted to accelerate. It records how the driver’s preferred acceleration differed from the automated response under particular driving conditions. Xu and colleagues propose Context-driven Personalized Adaptive Cruise Control (CoP-ACC), which learns recurring acceleration patterns from manual overtaking and shared-control overrides, predicts an appropriate pattern from the approaching maneuver’s context, and refines it into a continuous trajectory.1 ...

TL;DR for operators A materials team choosing a surrogate model for polymer screening should not treat molecular representation as a preprocessing detail. PolyBench261 compares language-model, graph-based, and descriptor-based approaches across 249,962 polymer-property datapoints, eight properties, four polymer architectures, and four evaluation tasks. Graph-based Polymer Chemprop (PC) and Polymer Periodic Graph (PPG) generally achieve the lowest errors on the evaluated in-distribution tasks. ...

TL;DR for operators A fixed seed is not a complete reproducibility guarantee. More surprisingly, neither is restoring the complete state of the random-number generator. In Reproducible AI Requires Reproducible Randomness, Anthony Bertrand, Tom Schmitt, Engelbert Mephu Nguifo, and David Hill compare library implementations of Mersenne Twister and Philox against canonical reference generators.1 Their deterministic tests show that the same nominal generator, initialized from the corresponding full state, can still emit a different sequence because libraries change what happens between stored state and returned output. ...

TL;DR for operators A multimodal dataset can contain the same amount of image, text, audio, or other modality-specific data yet produce different model performance depending on how often those modalities occur together. Marcus Ma and Shrikanth Narayanan isolate that distinction in Paired Multimodal Scaling Laws1, varying pairing while holding modality-specific data budgets fixed across three supervised environments. ...

TL;DR for operators A smartphone activity-recognition model can perform well during development and then degrade when deployed on a different dataset, user population, or sensor position. The natural response is to search for a domain-generalization technique that performs best across these changes. A 410,400-experiment benchmark suggests that this is the wrong unit of comparison. No individual objective, initialization strategy, or architectural modification wins consistently. Yet compatible combinations can produce materially larger gains: the best aggregate joint configurations improve accuracy by about 2.9 percentage points under cross-dataset shift and 4.9 points under cross-position shift. ...

TL;DR for operators A model reporting 99% accuracy creates an obvious decision pressure: should the team fund integration, start clinical workflow design, or treat the experiment as essentially solved? This paper is a useful example of why that decision cannot be made from the headline number alone. Its main results table reports a Stacking Classifier at 99.81% accuracy, Random Forest at 99.52%, and Bagging at 99.45%.1 But those results come from one public Kaggle dataset after the original 5,110 records—249 stroke-positive and 4,861 stroke-negative—were reduced to a balanced working dataset containing 249 observations in each class. ...

TL;DR for operators Many enterprise prediction systems already perform well on structured tables, but those tables often contain information that traditional pipelines treat as symbols rather than meaning: product categories, descriptions, labels, and domain-specific terminology. The CASE framework explores whether language-model-derived representations can add this missing semantic layer without replacing the tabular models already used in production. ...

TL;DR for operators A development team wants to identify energy-expensive Java methods early, without repeatedly profiling every code change on controlled hardware. Source-level features such as complexity, calls, loops, and control flow appear to offer a cheap signal because they can be extracted before the software runs. In this study, those features were almost useless without runtime evidence. When the researchers reran the same Random Forest prediction setup after removing execution time, predictive $R^2$ fell from 0.454 to 0.005. Code features that look operationally meaningful therefore provided almost no useful prediction until a timing signal was added. ...