LLM-Inspired New Generation Machine Learning: Hyperparameter Optimization and Uncertainty Quantification

Dear Colleagues,

Large Language Models (LLMs) have redefined scalability, context-awareness, and generalization in AI. Beyond text, their architectural innovations—massive parameterization, in-context learning, and implicit ensemble behavior—offer transformative insights for core machine learning challenges, particularly hyperparameter optimization (HPO) and uncertainty quantification (UQ). Traditional ML struggles with brittle HPO and poorly calibrated uncertainties, especially under data imbalance, distribution shift, or high-stakes deployment. This Special Issue seeks to bridge LLM-inspired paradigms with rigorous UQ and adaptive HPO to overcome these limitations.

We invite contributions that re-imagine Bayesian frameworks through an LLM lens. Key directions include:

• LLM-driven HPO: meta-learning of search spaces, transformer-based surrogate models, or prompt-guided Bayesian optimization;
• Well-calibrated UQ: Bayesian Model Averaging via ensemble-of-thoughts, Bayesian neural networks with attention-weighted priors, or test-time entropy regularization inspired by chain-of-thought sampling;
• Real-world robustness: UQ under class imbalance (e.g., rare disease detection), survival analysis with censored data, or financial time-series with non-stationarity.

Applications of interest span medicine (clinician confidence scoring, personalized survival curves), finance (liquidity risk forecasting, volatility UQ), drug design (protein synthesis), and business (dynamic pricing under demand uncertainty, supply-chain disruption modeling). Methods addressing out-of-distribution detection, adversarial robustness, or scalable inference are particularly encouraged.

Submissions may include original research, reproducible benchmarks, or perspective pieces linking LLM mechanisms (e.g., emergent ensembling, scaling laws) to UQ theory. All papers must provide empirical validation on public or proprietary datasets, with code release strongly recommended per MAKE reproducibility standards.

This Special Issue will catalyze a “new generation” of trustworthy ML, where LLM-inspired adaptability meets Bayesian rigor to deliver actionable, calibrated predictions in critical domains.

Dr. Vitaly Schetinin
Dr. Livija I. Jakaite
Prof. Dr. Jaroslaw Krzywanski
Guest Editors

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 250 words) can be sent to the Editorial Office for assessment.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-anonymized peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Machine Learning and Knowledge Extraction is an international peer-reviewed open access monthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 1800 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.

• LLM-inspired ML
• Bayesian learning methodology
• hyperparameter optimization
• uncertainty quantification
• imbalanced data
• survival analysis
• financial forecasting