About the Journal

The Journal of Research Hyperparameter Optimization (JRHO) is an open-access, peer-reviewed (double-blind) scientific journal that publishes high-quality research on hyperparameter optimization in machine learning and deep learning. JRHO serves as a global platform for researchers, industry practitioners, and policymakers to share theoretical and applied findings that enhance model performance, computational efficiency, and experiment reproducibility.

JRHO accepts manuscripts that explore methods, theories, algorithms, software, and real-world case studies related to how hyperparameters are selected, optimized, and managed throughout the entire ML lifecycle, from initial experiments, AutoML, to large-scale deployment. The journal also encourages open science practices (reproducible code, data, and protocols) and addresses ethical considerations, energy/carbon costs, and computational sustainability in the optimization process.

Scope

Topics (but not limited to) include:

  • Hyperparameter Optimization Methods: Bayesian optimization, SMBO/SMAC, TPE, Hyperband/SHA, BOHB, PBT, evolutionary & metaheuristic (GA, PSO, CMA-ES), bandit, simulated annealing, and hybrid approaches.
  • Hypergradient & Derivative-Based Optimization: Differentiable HPO, implicit differentiation, bilevel optimization.
  • Multi-Fidelity & Early-Stopping: Surrogate, learning curve learning, pruning/ASHA, low-fidelity proxy.
  • Multi-Objective & Constrained HPO: Accuracy-latency-memory-energy/cost; fairness & robustness as objectives.
  • Scalability & Systems: Distributed/parallel HPO, resource scheduling, caching, checkpointing, fault tolerance, HPO on GPU/TPU/edge.
  • AutoML & Pipeline-Level Tuning: Feature selection, preprocessing, model architecture, and end-to-end orchestration.
  • Neural Architecture Search (NAS) & Co-Tuning: NAS, one-shot/supernet, NAS+HPO combinations, low-cost NAS.
  • HPO for Large Models & Fine-Tuning: LLM/vision foundation models (LoRA, weight decay, lr schedule, batch size, quantization-aware tuning).
  • Theory & Guarantees: Generalization, sample complexity, objective landscape, convergence, uncertainty & calibration.
  • Robustness, Security & Privacy: HPO under adversarial attacks, DP-SGD, federated HPO, drift & continual learning.
  • Evaluation & Reproducibility: Benchmark design, protocols, variance reduction, seeding, metric reporting & uncertainty intervals.
  • Tools & Infrastructure: Libraries, frameworks, services, and MLOps practices for HPO in production.
  • Real-world Applications: Computer vision, NLP, tabular/financial, healthcare, recommendations, time series, robotics, and computational science.
  • Social & Environmental Impact: Energy costs/carbon footprint of HPO, algorithmic efficiency, and resource usage policies.

Publication Model

  • Access: Open access.

  • Peer Review: Double-blind.

  • Frequency: Biannual (two issues per year).

  • Language: English (optional translated abstracts).

  • Article Types: Research articles, short papers/letters, surveys/tutorials, systems & software papers, reproducibility reports.

Mission Statement

JRHO mission is to advance accurate, efficient, and reproducible hyperparameter optimization so that ML/DL systems are more accurate, resource-aware, fair, and production-ready.